October 8, 2026
2026-10-08
https://www.youtube.com/watch?v=iT-jky6EfJA

[0:23] We go live in 10.
[0:37] Every morning, the future moves faster.
[0:42] Every morning, the future moves faster. The models, the money, the power, and
[0:46] The models, the money, the power, and
[0:46] The models, the money, the power, and the people building what comes next.
[1:21] from Austin, Texas to the world.
[1:28] AI news, real context. This is the
[1:32] AI news, real context. This is the Austin Wilson show. I'm Austin Wilson,
[1:35] Austin Wilson show. I'm Austin Wilson,
[1:35] Austin Wilson show. I'm Austin Wilson, your human in the loop.
[1:42] Welcome to the Austin Wilson show where
[1:45] Welcome to the Austin Wilson show where we are breaking the AI headlines that
[1:47] we are breaking the AI headlines that
[1:47] we are breaking the AI headlines that matter to you. Today, my name is Austin
[1:50] matter to you. Today, my name is Austin
[1:50] matter to you. Today, my name is Austin Wilson. We're coming to you live from
[1:53] Wilson. We're coming to you live from
[1:53] Wilson. We're coming to you live from the AI capital of the world, Austin,
[1:55] the AI capital of the world, Austin,
[1:55] the AI capital of the world, Austin, Texas. Today, Chat GBT now answers with
[1:59] Texas. Today, Chat GBT now answers with
[1:59] Texas. Today, Chat GBT now answers with ready-made applications, not just text.
[2:02] ready-made applications, not just text.
[2:02] ready-made applications, not just text. Claude HiQ55
[2:04] Claude HiQ55
[2:04] Claude HiQ55 cuts costs 75%.
[2:08] cuts costs 75%.
[2:08] cuts costs 75%. Google lets you type to create a video
[2:10] Google lets you type to create a video
[2:10] Google lets you type to create a video game on the fly. Fired Open AI Trio
[2:14] game on the fly. Fired Open AI Trio
[2:14] game on the fly. Fired Open AI Trio pleads for auditable AI. Windows 11 now
[2:18] pleads for auditable AI. Windows 11 now
[2:18] pleads for auditable AI. Windows 11 now sandboxes AI agents and Google opens its
[2:23] sandboxes AI agents and Google opens its
[2:23] sandboxes AI agents and Google opens its synth ID AI checker to the public. Let's
[2:26] synth ID AI checker to the public. Let's
[2:26] synth ID AI checker to the public. Let's get into it. We got a heavy docket
[2:28] get into it. We got a heavy docket
[2:28] get into it. We got a heavy docket today. Excited to chat with you all. Uh
[2:32] today. Excited to chat with you all. Uh
[2:32] today. Excited to chat with you all. Uh chat GPT, go ahead and ask it how a
[2:35] chat GPT, go ahead and ask it how a
[2:35] chat GPT, go ahead and ask it how a bicycle works and instead of uh
[2:37] bicycle works and instead of uh
[2:38] bicycle works and instead of uh returning a bunch of text, it's going to
[2:39] returning a bunch of text, it's going to
[2:39] returning a bunch of text, it's going to create a clickable bicycle application
[2:42] create a clickable bicycle application
[2:42] create a clickable bicycle application that explains it to you. Um, what
[2:44] that explains it to you. Um, what
[2:44] that explains it to you. Um, what they're doing is replacing chat GPT56
[2:47] they're doing is replacing chat GPT56
[2:47] they're doing is replacing chat GPT56 with GPT6 in chat GPT's standard chat uh
[2:52] with GPT6 in chat GPT's standard chat uh
[2:52] with GPT6 in chat GPT's standard chat uh tab GPT6 soul for plus pro business and
[2:56] tab GPT6 soul for plus pro business and
[2:56] tab GPT6 soul for plus pro business and enterprise now uh in GPT6 Luna for free
[3:01] enterprise now uh in GPT6 Luna for free
[3:01] enterprise now uh in GPT6 Luna for free and go from today onward. Uh it ships
[3:05] and go from today onward. Uh it ships
[3:05] and go from today onward. Uh it ships with uh what they're calling intelligent
[3:07] with uh what they're calling intelligent
[3:08] with uh what they're calling intelligent UI. So, an intelligent user interface uh
[3:11] UI. So, an intelligent user interface uh
[3:11] UI. So, an intelligent user interface uh which answers with charts, maps,
[3:14] which answers with charts, maps,
[3:14] which answers with charts, maps, buttons, forms, and on the-fly
[3:17] buttons, forms, and on the-fly
[3:17] buttons, forms, and on the-fly calculators instead of just plain text.
[3:21] calculators instead of just plain text.
[3:21] calculators instead of just plain text. GPT6 hits every Chat GBT tiers chat tab
[3:26] GPT6 hits every Chat GBT tiers chat tab
[3:26] GPT6 hits every Chat GBT tiers chat tab with interactive visual answers. So
[3:28] with interactive visual answers. So
[3:28] with interactive visual answers. So whether you're uh using chat GPT for
[3:32] whether you're uh using chat GPT for
[3:32] whether you're uh using chat GPT for free or paying for the top model uh you
[3:34] free or paying for the top model uh you
[3:34] free or paying for the top model uh you are going to uh now see chat GPT6
[3:37] are going to uh now see chat GPT6
[3:38] are going to uh now see chat GPT6 available to you which is the most
[3:39] available to you which is the most
[3:39] available to you which is the most powerful uh model that they have
[3:41] powerful uh model that they have
[3:42] powerful uh model that they have available. This is the default chat GBT
[3:44] available. This is the default chat GBT
[3:44] available. This is the default chat GBT that one 1.2 two billion weekly users
[3:48] that one 1.2 two billion weekly users
[3:48] that one 1.2 two billion weekly users touch and it shifts the chatbot from uh
[3:51] touch and it shifts the chatbot from uh
[3:52] touch and it shifts the chatbot from uh just a text box to an interface that
[3:54] just a text box to an interface that
[3:54] just a text box to an interface that builds itself around your question work
[3:57] builds itself around your question work
[3:57] builds itself around your question work and codeex models are unchanged. Uh
[4:00] and codeex models are unchanged. Uh
[4:00] and codeex models are unchanged. Uh OpenAI says GPT6 with intelligent UI
[4:04] OpenAI says GPT6 with intelligent UI
[4:04] OpenAI says GPT6 with intelligent UI makes everyday answers visual and
[4:07] makes everyday answers visual and
[4:07] makes everyday answers visual and interactive for their over 1 billion
[4:10] interactive for their over 1 billion
[4:10] interactive for their over 1 billion weekly users. Free and Go are included
[4:13] weekly users. Free and Go are included
[4:13] weekly users. Free and Go are included in this of course. Demos show a
[4:15] in this of course. Demos show a
[4:15] in this of course. Demos show a clickable seven-speed bike, a recipe
[4:18] clickable seven-speed bike, a recipe
[4:18] clickable seven-speed bike, a recipe that rescales by guest count, a bill
[4:21] that rescales by guest count, a bill
[4:22] that rescales by guest count, a bill splitter, and a hiking map. Uh, GPT6
[4:26] splitter, and a hiking map. Uh, GPT6
[4:26] splitter, and a hiking map. Uh, GPT6 instant starts web answers 44% sooner.
[4:30] instant starts web answers 44% sooner.
[4:30] instant starts web answers 44% sooner. Uh, GPT6 isn't brand new. It hit work
[4:33] Uh, GPT6 isn't brand new. It hit work
[4:33] Uh, GPT6 isn't brand new. It hit work and codeex applications on September
[4:36] and codeex applications on September
[4:36] and codeex applications on September 22nd and GPT 6.1 Soul shipped on
[4:40] 22nd and GPT 6.1 Soul shipped on
[4:40] 22nd and GPT 6.1 Soul shipped on September 29th. The big difference is uh
[4:43] September 29th. The big difference is uh
[4:43] September 29th. The big difference is uh all of the uh different subscription
[4:45] all of the uh different subscription
[4:45] all of the uh different subscription tiers and the free users get access to
[4:48] tiers and the free users get access to
[4:48] tiers and the free users get access to it now.
[4:50] it now.
[4:50] it now. So uh you know this is really
[4:51] So uh you know this is really
[4:52] So uh you know this is really interesting. Uh you know ask it to split
[4:54] interesting. Uh you know ask it to split
[4:54] interesting. Uh you know ask it to split your your tab at um dinner and it's
[4:57] your your tab at um dinner and it's
[4:57] your your tab at um dinner and it's going to uh automatically create an
[5:00] going to uh automatically create an
[5:00] going to uh automatically create an application that can do that for you. Uh
[5:02] application that can do that for you. Uh
[5:02] application that can do that for you. Uh the funny thing is two years ago I spent
[5:04] the funny thing is two years ago I spent
[5:04] the funny thing is two years ago I spent about a month with uh very nent uh um AI
[5:12] about a month with uh very nent uh um AI
[5:12] about a month with uh very nent uh um AI model. So you know chat GPT4 must have
[5:15] model. So you know chat GPT4 must have
[5:15] model. So you know chat GPT4 must have been at the time. Uh and it took me a
[5:18] been at the time. Uh and it took me a
[5:18] been at the time. Uh and it took me a month at to build a apps a uh receipt
[5:21] month at to build a apps a uh receipt
[5:21] month at to build a apps a uh receipt splitter application. So it's really
[5:23] splitter application. So it's really
[5:23] splitter application. So it's really funny. Now, uh you can um text chat GPT
[5:27] funny. Now, uh you can um text chat GPT
[5:27] funny. Now, uh you can um text chat GPT in a few words and it's going to create
[5:29] in a few words and it's going to create
[5:29] in a few words and it's going to create that for you on the fly, which is
[5:31] that for you on the fly, which is
[5:31] that for you on the fly, which is amazing to see that progress in just
[5:33] amazing to see that progress in just
[5:33] amazing to see that progress in just under uh 2 years. Um now, I think people
[5:38] under uh 2 years. Um now, I think people
[5:38] under uh 2 years. Um now, I think people are going to love this kind of stuff
[5:39] are going to love this kind of stuff
[5:39] are going to love this kind of stuff from a form faction or form factor type
[5:43] from a form faction or form factor type
[5:43] from a form faction or form factor type usable functional way to interact with
[5:46] usable functional way to interact with
[5:46] usable functional way to interact with chat GBT. Uh typically we see uh on
[5:50] chat GBT. Uh typically we see uh on
[5:50] chat GBT. Uh typically we see uh on these AI applications, they're extremely
[5:52] these AI applications, they're extremely
[5:52] these AI applications, they're extremely verbose. We get, you know, paragraphs
[5:55] verbose. We get, you know, paragraphs
[5:55] verbose. We get, you know, paragraphs and paragraphs of text. It's like I
[5:57] and paragraphs of text. It's like I
[5:58] and paragraphs of text. It's like I don't I don't want to read that whole
[5:59] don't I don't want to read that whole
[5:59] don't I don't want to read that whole thing. I don't have time to read that.
[6:01] thing. I don't have time to read that.
[6:01] thing. I don't have time to read that. Just I want an answer to my question, a
[6:03] Just I want an answer to my question, a
[6:03] Just I want an answer to my question, a simple explanation. Sometimes you have
[6:05] simple explanation. Sometimes you have
[6:05] simple explanation. Sometimes you have to instruct it. Just give me give it
[6:07] to instruct it. Just give me give it
[6:07] to instruct it. Just give me give it give it to me in three bullet points,
[6:08] give it to me in three bullet points,
[6:08] give it to me in three bullet points, you know. So, it'll be interesting to
[6:10] you know. So, it'll be interesting to
[6:10] you know. So, it'll be interesting to see on these like functional
[6:13] see on these like functional
[6:13] see on these like functional um useful applications having it produce
[6:16] um useful applications having it produce
[6:16] um useful applications having it produce an actual like interactive application
[6:19] an actual like interactive application
[6:20] an actual like interactive application that that you can use versus spitting
[6:22] that that you can use versus spitting
[6:22] that that you can use versus spitting out uh text. Um I'm I'm maybe a little
[6:27] out uh text. Um I'm I'm maybe a little
[6:27] out uh text. Um I'm I'm maybe a little bit skept skeptical on um how extensible
[6:30] bit skept skeptical on um how extensible
[6:30] bit skept skeptical on um how extensible it is. So, are people actually going to
[6:33] it is. So, are people actually going to
[6:33] it is. So, are people actually going to use and create these applications? And
[6:36] use and create these applications? And
[6:36] use and create these applications? And uh is it going to have much of an
[6:37] uh is it going to have much of an
[6:37] uh is it going to have much of an impact? Um, you tell me. Would you
[6:40] impact? Um, you tell me. Would you
[6:40] impact? Um, you tell me. Would you rather your chatbot hand you uh a mini
[6:43] rather your chatbot hand you uh a mini
[6:43] rather your chatbot hand you uh a mini app or uh just give you the answer in a
[6:48] app or uh just give you the answer in a
[6:48] app or uh just give you the answer in a sentence. Curious to hear from our
[6:50] sentence. Curious to hear from our
[6:50] sentence. Curious to hear from our audience on that. Uh now, Anthropic just
[6:54] audience on that. Uh now, Anthropic just
[6:54] audience on that. Uh now, Anthropic just made its small model about 75% cheaper
[6:56] made its small model about 75% cheaper
[6:56] made its small model about 75% cheaper to run and priced it head-to-head with
[6:59] to run and priced it head-to-head with
[6:59] to run and priced it head-to-head with OpenAI's cheaper model GPT6 Luna.
[7:03] OpenAI's cheaper model GPT6 Luna.
[7:03] OpenAI's cheaper model GPT6 Luna. Anthropic released Claude Haiku 55 on
[7:06] Anthropic released Claude Haiku 55 on
[7:06] Anthropic released Claude Haiku 55 on October 7th.
[7:08] October 7th.
[7:08] October 7th. uh for uh 10 cents per million input
[7:11] uh for uh 10 cents per million input
[7:11] uh for uh 10 cents per million input tokens, which is pretty phenomenal.
[7:13] tokens, which is pretty phenomenal.
[7:13] tokens, which is pretty phenomenal. That's 90% below uh their previous haiku
[7:17] That's 90% below uh their previous haiku
[7:17] That's 90% below uh their previous haiku model. Um it's the first haiku with an
[7:19] model. Um it's the first haiku with an
[7:19] model. Um it's the first haiku with an adjustable effort setting. Uh Anthropic
[7:22] adjustable effort setting. Uh Anthropic
[7:22] adjustable effort setting. Uh Anthropic also cut in half their Sonnet 55 cash
[7:26] also cut in half their Sonnet 55 cash
[7:26] also cut in half their Sonnet 55 cash read prices and added monthly API
[7:28] read prices and added monthly API
[7:28] read prices and added monthly API credits for max and team subscribers.
[7:31] credits for max and team subscribers.
[7:31] credits for max and team subscribers. Cool stuff for uh the nerds. Uh we love
[7:36] Cool stuff for uh the nerds. Uh we love
[7:36] Cool stuff for uh the nerds. Uh we love it. Okay, Anthropic's new small model
[7:39] it. Okay, Anthropic's new small model
[7:39] it. Okay, Anthropic's new small model cost about 75% less than Haiku 45 and
[7:42] cost about 75% less than Haiku 45 and
[7:42] cost about 75% less than Haiku 45 and lands at uh GPT6 Luna's prices. Small
[7:45] lands at uh GPT6 Luna's prices. Small
[7:45] lands at uh GPT6 Luna's prices. Small model prices just fell another order of
[7:47] model prices just fell another order of
[7:47] model prices just fell another order of magnitude and Anthropic and Open III now
[7:50] magnitude and Anthropic and Open III now
[7:50] magnitude and Anthropic and Open III now charge the same base rate. So the cheap
[7:52] charge the same base rate. So the cheap
[7:52] charge the same base rate. So the cheap tier is turning into uh straight price
[7:54] tier is turning into uh straight price
[7:54] tier is turning into uh straight price warp. Um Anthropic calls Haiku the
[7:57] warp. Um Anthropic calls Haiku the
[7:57] warp. Um Anthropic calls Haiku the cheapest, fastest, and most capable
[7:59] cheapest, fastest, and most capable
[7:59] cheapest, fastest, and most capable small model ever released. Um, Anthropic
[8:03] small model ever released. Um, Anthropic
[8:03] small model ever released. Um, Anthropic reported that the the model performed
[8:06] reported that the the model performed
[8:06] reported that the the model performed exceptionally well on their internal
[8:08] exceptionally well on their internal
[8:08] exceptionally well on their internal benchmarks. That's yet to be uh tested
[8:11] benchmarks. That's yet to be uh tested
[8:11] benchmarks. That's yet to be uh tested by third parties. So, we'll see where
[8:13] by third parties. So, we'll see where
[8:13] by third parties. So, we'll see where that actually lands. Um, but but great
[8:16] that actually lands. Um, but but great
[8:16] that actually lands. Um, but but great to see Anthropic releasing uh this
[8:19] to see Anthropic releasing uh this
[8:19] to see Anthropic releasing uh this latest update. Uh, what does this mean
[8:22] latest update. Uh, what does this mean
[8:22] latest update. Uh, what does this mean for for us? It it means that continually
[8:25] for for us? It it means that continually
[8:25] for for us? It it means that continually we're seeing the intelligence improve
[8:28] we're seeing the intelligence improve
[8:28] we're seeing the intelligence improve and and improve at a substantial rate.
[8:31] and and improve at a substantial rate.
[8:31] and and improve at a substantial rate. uh and the cost per intelligence or cost
[8:34] uh and the cost per intelligence or cost
[8:34] uh and the cost per intelligence or cost per token is dropping significantly. So
[8:37] per token is dropping significantly. So
[8:37] per token is dropping significantly. So that means that um as
[8:40] that means that um as
[8:40] that means that um as you know more and more complex tasks are
[8:43] you know more and more complex tasks are
[8:43] you know more and more complex tasks are capable we you know chat GPT um you know
[8:47] capable we you know chat GPT um you know
[8:47] capable we you know chat GPT um you know providing solutions to all of these math
[8:49] providing solutions to all of these math
[8:49] providing solutions to all of these math problems uh yesterday
[8:52] problems uh yesterday
[8:52] problems uh yesterday um as as kind of the frontier the the
[8:55] um as as kind of the frontier the the
[8:55] um as as kind of the frontier the the edge intelligence increases and improves
[8:57] edge intelligence increases and improves
[8:57] edge intelligence increases and improves and does more and more difficult things
[9:00] and does more and more difficult things
[9:00] and does more and more difficult things uh these companies are figuring out how
[9:02] uh these companies are figuring out how
[9:02] uh these companies are figuring out how to make you know still really smart uh
[9:06] to make you know still really smart uh
[9:06] to make you know still really smart uh really capable models make them less
[9:08] really capable models make them less
[9:08] really capable models make them less expensive and make them smarter and
[9:11] expensive and make them smarter and
[9:11] expensive and make them smarter and faster. Um, and so the the speed and uh
[9:15] faster. Um, and so the the speed and uh
[9:15] faster. Um, and so the the speed and uh the lower cost is amazing uh for the
[9:18] the lower cost is amazing uh for the
[9:18] the lower cost is amazing uh for the everyday use user. We probably, you
[9:22] everyday use user. We probably, you
[9:22] everyday use user. We probably, you know, if if you're asking questions of
[9:25] know, if if you're asking questions of
[9:25] know, if if you're asking questions of Claude, uh, using the new Haiku 55
[9:28] Claude, uh, using the new Haiku 55
[9:28] Claude, uh, using the new Haiku 55 model,
[9:30] model,
[9:30] model, like on a regular plan, you would
[9:32] like on a regular plan, you would
[9:32] like on a regular plan, you would probably never ever in a month, um, hit
[9:36] probably never ever in a month, um, hit
[9:36] probably never ever in a month, um, hit your max, you usage limit. If you're
[9:39] your max, you usage limit. If you're
[9:39] your max, you usage limit. If you're just asking it questions, you know, hey,
[9:41] just asking it questions, you know, hey,
[9:41] just asking it questions, you know, hey, help me with, uh, you know, presentation
[9:43] help me with, uh, you know, presentation
[9:43] help me with, uh, you know, presentation I have coming up. Uh, help me, you know,
[9:47] I have coming up. Uh, help me, you know,
[9:47] I have coming up. Uh, help me, you know, make a recipe for dinner tonight. like
[9:48] make a recipe for dinner tonight. like
[9:48] make a recipe for dinner tonight. like that kind of stuff is just going to be
[9:50] that kind of stuff is just going to be
[9:50] that kind of stuff is just going to be such a small load on these types of
[9:53] such a small load on these types of
[9:53] such a small load on these types of models that you would never even get
[9:55] models that you would never even get
[9:55] models that you would never even get close to hitting any sort of limit with
[9:57] close to hitting any sort of limit with
[9:57] close to hitting any sort of limit with that. What's interesting as we scale
[9:59] that. What's interesting as we scale
[9:59] that. What's interesting as we scale this up and the reason this matters is
[10:02] this up and the reason this matters is
[10:02] this up and the reason this matters is um we're seeing the surface area of
[10:04] um we're seeing the surface area of
[10:04] um we're seeing the surface area of intelligence increase across the board,
[10:06] intelligence increase across the board,
[10:06] intelligence increase across the board, right? What does that mean? So, it means
[10:08] right? What does that mean? So, it means
[10:08] right? What does that mean? So, it means not just users
[10:11] not just users
[10:11] not just users using chat GPT or claude on their phone.
[10:14] using chat GPT or claude on their phone.
[10:14] using chat GPT or claude on their phone. It means that um everything that we
[10:17] It means that um everything that we
[10:17] It means that um everything that we touch will have some intelligence built
[10:19] touch will have some intelligence built
[10:19] touch will have some intelligence built into it. And so when we have these um
[10:23] into it. And so when we have these um
[10:23] into it. And so when we have these um you know really smart but really
[10:25] you know really smart but really
[10:25] you know really smart but really efficient models uh that it means that
[10:28] efficient models uh that it means that
[10:28] efficient models uh that it means that that can proliferate in all our devices,
[10:31] that can proliferate in all our devices,
[10:31] that can proliferate in all our devices, all our tools become smarter and
[10:33] all our tools become smarter and
[10:33] all our tools become smarter and smarter. Especially when you talk about,
[10:34] smarter. Especially when you talk about,
[10:34] smarter. Especially when you talk about, you know, the kind of obvious
[10:35] you know, the kind of obvious
[10:35] you know, the kind of obvious application is talking about robots in
[10:38] application is talking about robots in
[10:38] application is talking about robots in our day-to-day and the robots need
[10:39] our day-to-day and the robots need
[10:39] our day-to-day and the robots need intelligence uh to to uh run and operate
[10:43] intelligence uh to to uh run and operate
[10:43] intelligence uh to to uh run and operate properly uh and being able to run those
[10:46] properly uh and being able to run those
[10:46] properly uh and being able to run those more efficiently both from a speed
[10:48] more efficiently both from a speed
[10:48] more efficiently both from a speed standpoint and a cost standpoint just
[10:49] standpoint and a cost standpoint just
[10:49] standpoint and a cost standpoint just makes those uh those types of machines
[10:51] makes those uh those types of machines
[10:51] makes those uh those types of machines work better. So, it'll be interesting um
[10:55] work better. So, it'll be interesting um
[10:55] work better. So, it'll be interesting um as like we continue to follow this trend
[10:58] as like we continue to follow this trend
[10:58] as like we continue to follow this trend and this just pace of um improvement is
[11:02] and this just pace of um improvement is
[11:02] and this just pace of um improvement is uh phenomenal to watch as well too. So,
[11:04] uh phenomenal to watch as well too. So,
[11:04] uh phenomenal to watch as well too. So, all of the above. Um congrats to
[11:07] all of the above. Um congrats to
[11:07] all of the above. Um congrats to Anthropic. Excited about uh you know,
[11:11] Anthropic. Excited about uh you know,
[11:12] Anthropic. Excited about uh you know, better intelligence, lower cost. Um,
[11:14] better intelligence, lower cost. Um,
[11:14] better intelligence, lower cost. Um, I'll ask you guys if uh GPT6, Luna, and
[11:18] I'll ask you guys if uh GPT6, Luna, and
[11:18] I'll ask you guys if uh GPT6, Luna, and Haiku cost the same, what makes you pick
[11:21] Haiku cost the same, what makes you pick
[11:21] Haiku cost the same, what makes you pick one o over the other? So, uh, if you're
[11:24] one o over the other? So, uh, if you're
[11:24] one o over the other? So, uh, if you're going between chat, GPT, and Claude, why
[11:27] going between chat, GPT, and Claude, why
[11:27] going between chat, GPT, and Claude, why do you choose one over the other?
[11:28] do you choose one over the other?
[11:28] do you choose one over the other? Curious to hear your thoughts on the
[11:31] Curious to hear your thoughts on the
[11:31] Curious to hear your thoughts on the matter. Google Labs new playground turns
[11:35] matter. Google Labs new playground turns
[11:35] matter. Google Labs new playground turns a text prompt into a playable browser
[11:37] a text prompt into a playable browser
[11:38] a text prompt into a playable browser game. No code and only for US adults, 18
[11:41] game. No code and only for US adults, 18
[11:41] game. No code and only for US adults, 18 plus right now. Uh Google Labs launched
[11:44] plus right now. Uh Google Labs launched
[11:44] plus right now. Uh Google Labs launched Playground on Wednesday at
[11:45] Playground on Wednesday at
[11:45] Playground on Wednesday at playground.google.
[11:48] playground.google.
[11:48] playground.google. Uh you describe a game. Pick 2D or 3D
[11:51] Uh you describe a game. Pick 2D or 3D
[11:51] Uh you describe a game. Pick 2D or 3D and single player or multiplayer and uh
[11:53] and single player or multiplayer and uh
[11:53] and single player or multiplayer and uh just keep chatting to change the rules,
[11:55] just keep chatting to change the rules,
[11:55] just keep chatting to change the rules, physics or characters. Games run in a
[11:57] physics or characters. Games run in a
[11:58] physics or characters. Games run in a browser on phone or laptop and can be
[12:01] browser on phone or laptop and can be
[12:01] browser on phone or laptop and can be kept private or you could share them
[12:03] kept private or you could share them
[12:03] kept private or you could share them with your friends or publish them to uh
[12:06] with your friends or publish them to uh
[12:06] with your friends or publish them to uh a gallery which is really fun. This
[12:08] a gallery which is really fun. This
[12:08] a gallery which is really fun. This sounds awesome. Uh Google's new
[12:10] sounds awesome. Uh Google's new
[12:10] sounds awesome. Uh Google's new playground builds browser games from
[12:12] playground builds browser games from
[12:12] playground builds browser games from text prompts. Uh US adults 18 plus only.
[12:16] text prompts. Uh US adults 18 plus only.
[12:16] text prompts. Uh US adults 18 plus only. Make a game. Um
[12:18] Make a game. Um
[12:18] Make a game. Um making games used to take you know
[12:20] making games used to take you know
[12:20] making games used to take you know hours, weeks, months, years of coding to
[12:23] hours, weeks, months, years of coding to
[12:23] hours, weeks, months, years of coding to develop and build these games and now
[12:25] develop and build these games and now
[12:25] develop and build these games and now anyone can make one just by typing. Um
[12:28] anyone can make one just by typing. Um
[12:28] anyone can make one just by typing. Um and this puts Google right up against
[12:30] and this puts Google right up against
[12:30] and this puts Google right up against Roblox's a uh AI builder and Unity speak
[12:34] Roblox's a uh AI builder and Unity speak
[12:34] Roblox's a uh AI builder and Unity speak uh integration which is uh coming next.
[12:38] uh integration which is uh coming next.
[12:38] uh integration which is uh coming next. Google says anyone can create, play, and
[12:40] Google says anyone can create, play, and
[12:40] Google says anyone can create, play, and share games by typing with uh no quote
[12:44] share games by typing with uh no quote
[12:44] share games by typing with uh no quote coding required. Google's own blog post
[12:47] coding required. Google's own blog post
[12:47] coding required. Google's own blog post up um and TechCrunch which notes free
[12:51] up um and TechCrunch which notes free
[12:51] up um and TechCrunch which notes free weekly tokens and higher limits for
[12:54] weekly tokens and higher limits for
[12:54] weekly tokens and higher limits for Google one uh AI subscribers. So
[12:57] Google one uh AI subscribers. So
[12:57] Google one uh AI subscribers. So multiple uh different confirmations of
[13:00] multiple uh different confirmations of
[13:00] multiple uh different confirmations of this story experimental and uh US 18
[13:04] this story experimental and uh US 18
[13:04] this story experimental and uh US 18 plus only. Google hasn't said how many
[13:06] plus only. Google hasn't said how many
[13:06] plus only. Google hasn't said how many tokens you get and there's no
[13:07] tokens you get and there's no
[13:07] tokens you get and there's no independent tests of how good the games
[13:09] independent tests of how good the games
[13:09] independent tests of how good the games are, but I guarantee we're going to see
[13:11] are, but I guarantee we're going to see
[13:11] are, but I guarantee we're going to see some awesome games come out of this uh
[13:13] some awesome games come out of this uh
[13:14] some awesome games come out of this uh in the coming days and weeks. Uh really
[13:17] in the coming days and weeks. Uh really
[13:17] in the coming days and weeks. Uh really fun advancement kind of interesting. You
[13:19] fun advancement kind of interesting. You
[13:19] fun advancement kind of interesting. You know, we're seeing
[13:21] know, we're seeing
[13:21] know, we're seeing uh progress uh across the board week
[13:25] uh progress uh across the board week
[13:25] uh progress uh across the board week over week. Uh and and even it's like it
[13:28] over week. Uh and and even it's like it
[13:28] over week. Uh and and even it's like it gets compressed down to days now. um you
[13:32] gets compressed down to days now. um you
[13:32] gets compressed down to days now. um you know just last week Google launching
[13:33] know just last week Google launching
[13:33] know just last week Google launching their uh you know Frontier model Argon
[13:37] their uh you know Frontier model Argon
[13:37] their uh you know Frontier model Argon or or at least announcing that it's not
[13:40] or or at least announcing that it's not
[13:40] or or at least announcing that it's not released yet and then continue to see
[13:42] released yet and then continue to see
[13:42] released yet and then continue to see more and more developments out of Google
[13:43] more and more developments out of Google
[13:44] more and more developments out of Google which is exciting. This one is fun. Um,
[13:46] which is exciting. This one is fun. Um,
[13:46] which is exciting. This one is fun. Um, that being said, uh, you you've been
[13:50] that being said, uh, you you've been
[13:50] that being said, uh, you you've been able to do this for a long time with,
[13:52] able to do this for a long time with,
[13:52] able to do this for a long time with, you know, I remember m maybe o over a
[13:54] you know, I remember m maybe o over a
[13:54] you know, I remember m maybe o over a year ago and and simple versions, they
[13:57] year ago and and simple versions, they
[13:57] year ago and and simple versions, they don't work super great. But I remember a
[13:58] don't work super great. But I remember a
[13:58] don't work super great. But I remember a year ago in Claude saying, you know,
[14:01] year ago in Claude saying, you know,
[14:01] year ago in Claude saying, you know, make me uh, you know, a snake game and
[14:04] make me uh, you know, a snake game and
[14:04] make me uh, you know, a snake game and it just, you know, made it on the fly
[14:06] it just, you know, made it on the fly
[14:06] it just, you know, made it on the fly and it was completely playable. Very
[14:08] and it was completely playable. Very
[14:08] and it was completely playable. Very simple, only 2D. We're not talking about
[14:10] simple, only 2D. We're not talking about
[14:10] simple, only 2D. We're not talking about 3D stuff. the the the advancements in um
[14:15] 3D stuff. the the the advancements in um
[14:15] 3D stuff. the the the advancements in um game development, AI game development,
[14:18] game development, AI game development,
[14:18] game development, AI game development, and seeing all this um happen. You know,
[14:21] and seeing all this um happen. You know,
[14:21] and seeing all this um happen. You know, shout out to uh Peter Levels on on X and
[14:25] shout out to uh Peter Levels on on X and
[14:25] shout out to uh Peter Levels on on X and his competition um for uh his Vibe Jam
[14:30] his competition um for uh his Vibe Jam
[14:30] his competition um for uh his Vibe Jam competition building out uh Vivecoded
[14:33] competition building out uh Vivecoded
[14:34] competition building out uh Vivecoded video games. um they they've been doing
[14:36] video games. um they they've been doing
[14:36] video games. um they they've been doing that for a couple years. So, it's not
[14:38] that for a couple years. So, it's not
[14:38] that for a couple years. So, it's not completely new technology, but maybe the
[14:41] completely new technology, but maybe the
[14:41] completely new technology, but maybe the interesting thing to this is opening
[14:42] interesting thing to this is opening
[14:42] interesting thing to this is opening this up to the general public who's
[14:45] this up to the general public who's
[14:45] this up to the general public who's going to love to sit down and say, "Hey,
[14:47] going to love to sit down and say, "Hey,
[14:47] going to love to sit down and say, "Hey, um you know what? Are we you guys want
[14:50] um you know what? Are we you guys want
[14:50] um you know what? Are we you guys want to play board game tonight after
[14:51] to play board game tonight after
[14:51] to play board game tonight after dinner?" It's like, "Oh, no. Jimmy uh
[14:54] dinner?" It's like, "Oh, no. Jimmy uh
[14:54] dinner?" It's like, "Oh, no. Jimmy uh coded a new video game that we're going
[14:55] coded a new video game that we're going
[14:55] coded a new video game that we're going to play together as a family." Maybe
[14:57] to play together as a family." Maybe
[14:57] to play together as a family." Maybe that's a maybe that's a use case. I
[14:58] that's a maybe that's a use case. I
[14:58] that's a maybe that's a use case. I don't know. We'll see. Um but uh fun fun
[15:01] don't know. We'll see. Um but uh fun fun
[15:01] don't know. We'll see. Um but uh fun fun to see that. fun to see new developments
[15:03] to see that. fun to see new developments
[15:03] to see that. fun to see new developments from Google, whatever it is. Um,
[15:06] from Google, whatever it is. Um,
[15:06] from Google, whatever it is. Um, interesting, you know, be interesting to
[15:07] interesting, you know, be interesting to
[15:08] interesting, you know, be interesting to see what the capabilities are. Um, like
[15:10] see what the capabilities are. Um, like
[15:10] see what the capabilities are. Um, like I said, and um, yeah, congrats. That's
[15:13] I said, and um, yeah, congrats. That's
[15:13] I said, and um, yeah, congrats. That's fun. Uh, if you guys go to
[15:17] fun. Uh, if you guys go to
[15:17] fun. Uh, if you guys go to um, uh, to to to check this out, I'm
[15:20] um, uh, to to to check this out, I'm
[15:20] um, uh, to to to check this out, I'm curious to see what games that that you
[15:22] curious to see what games that that you
[15:22] curious to see what games that that you guys would play. Um, and if they're
[15:25] guys would play. Um, and if they're
[15:25] guys would play. Um, and if they're sharable, put them in the comments.
[15:28] sharable, put them in the comments.
[15:28] sharable, put them in the comments. Let's see. I would love to jump in and
[15:29] Let's see. I would love to jump in and
[15:29] Let's see. I would love to jump in and see what you guys are building. That
[15:30] see what you guys are building. That
[15:30] see what you guys are building. That sounds like a lot of fun.
[15:33] sounds like a lot of fun.
[15:33] sounds like a lot of fun. Oh, now we get into it. Three safety
[15:35] Oh, now we get into it. Three safety
[15:35] Oh, now we get into it. Three safety researchers Open AI fired last week are
[15:38] researchers Open AI fired last week are
[15:38] researchers Open AI fired last week are now telling its board to stop building
[15:41] now telling its board to stop building
[15:41] now telling its board to stop building AI it can't monitor.
[15:44] AI it can't monitor.
[15:44] AI it can't monitor. Where have we heard this before? The
[15:46] Where have we heard this before? The
[15:46] Where have we heard this before? The Wall Street Journal reports that Tom
[15:48] Wall Street Journal reports that Tom
[15:48] Wall Street Journal reports that Tom Carbach, Makita Blesny, and Jasmine
[15:51] Carbach, Makita Blesny, and Jasmine
[15:51] Carbach, Makita Blesny, and Jasmine Wang, three safety researchers OpenAI
[15:53] Wang, three safety researchers OpenAI
[15:54] Wang, three safety researchers OpenAI fired around October 1st over alleged
[15:55] fired around October 1st over alleged
[15:56] fired around October 1st over alleged mishandling of sensitive information,
[15:58] mishandling of sensitive information,
[15:58] mishandling of sensitive information, sent a letter to OpenAI's board and
[16:02] sent a letter to OpenAI's board and
[16:02] sent a letter to OpenAI's board and safety committees. The letter urges
[16:04] safety committees. The letter urges
[16:04] safety committees. The letter urges OpenAI and its rivals not to build
[16:07] OpenAI and its rivals not to build
[16:07] OpenAI and its rivals not to build models whose chain of thought reasoning
[16:09] models whose chain of thought reasoning
[16:09] models whose chain of thought reasoning can't be monitored. It says the fringes
[16:13] can't be monitored. It says the fringes
[16:13] can't be monitored. It says the fringes uh created a quote unquote chilling
[16:17] uh created a quote unquote chilling
[16:17] uh created a quote unquote chilling atmosphere
[16:18] atmosphere
[16:18] atmosphere uh for staff who remain. I'm sorry. It
[16:23] uh for staff who remain. I'm sorry. It
[16:23] uh for staff who remain. I'm sorry. It says the firings created a chilling
[16:25] says the firings created a chilling
[16:25] says the firings created a chilling atmosphere for staff who remain which I
[16:29] atmosphere for staff who remain which I
[16:29] atmosphere for staff who remain which I guess that's hard, you know. I guess
[16:31] guess that's hard, you know. I guess
[16:31] guess that's hard, you know. I guess yeah, natural reaction from getting
[16:33] yeah, natural reaction from getting
[16:33] yeah, natural reaction from getting fired. fired uh open AI safety
[16:36] fired. fired uh open AI safety
[16:36] fired. fired uh open AI safety researchers asked the board to stop
[16:38] researchers asked the board to stop
[16:38] researchers asked the board to stop building AI whose reasoning can't be
[16:42] building AI whose reasoning can't be
[16:42] building AI whose reasoning can't be monitored. Chain of thought monitoring
[16:44] monitored. Chain of thought monitoring
[16:44] monitored. Chain of thought monitoring is one of the few practical ways uh to
[16:48] is one of the few practical ways uh to
[16:48] is one of the few practical ways uh to see what a model is planning and
[16:50] see what a model is planning and
[16:50] see what a model is planning and OpenAI's own system card says its
[16:53] OpenAI's own system card says its
[16:53] OpenAI's own system card says its asterass models could evade those
[16:56] asterass models could evade those
[16:56] asterass models could evade those monitors. Um the the letter per Wall
[16:59] monitors. Um the the letter per Wall
[17:00] monitors. Um the the letter per Wall Street Journal says uh quote unquote as
[17:02] Street Journal says uh quote unquote as
[17:02] Street Journal says uh quote unquote as an industry we do not yet know how to
[17:04] an industry we do not yet know how to
[17:04] an industry we do not yet know how to safely develop and deploy models that we
[17:07] safely develop and deploy models that we
[17:07] safely develop and deploy models that we cannot monitor. Uh this is coming from
[17:10] cannot monitor. Uh this is coming from
[17:10] cannot monitor. Uh this is coming from Wall Street Journal. Gizmodo cites this
[17:14] Wall Street Journal. Gizmodo cites this
[17:14] Wall Street Journal. Gizmodo cites this um says Astroclass models could could
[17:17] um says Astroclass models could could
[17:17] um says Astroclass models could could evade uh chain of thought monitors uh
[17:20] evade uh chain of thought monitors uh
[17:20] evade uh chain of thought monitors uh under adversarial conditions. Uh, OpenAI
[17:24] under adversarial conditions. Uh, OpenAI
[17:24] under adversarial conditions. Uh, OpenAI says the firings were about data
[17:25] says the firings were about data
[17:25] says the firings were about data handling, not speaking out, and an
[17:27] handling, not speaking out, and an
[17:27] handling, not speaking out, and an internal memo says it strongly agreed
[17:30] internal memo says it strongly agreed
[17:30] internal memo says it strongly agreed with the recommendation. The full letter
[17:32] with the recommendation. The full letter
[17:32] with the recommendation. The full letter isn't public. Um,
[17:35] isn't public. Um,
[17:35] isn't public. Um, this is just uh, man, we see
[17:38] this is just uh, man, we see
[17:38] this is just uh, man, we see advancements week over week and then I
[17:40] advancements week over week and then I
[17:40] advancements week over week and then I feel like we have a safety story
[17:42] feel like we have a safety story
[17:42] feel like we have a safety story regarding Open AI on a weekly basis. I
[17:46] regarding Open AI on a weekly basis. I
[17:46] regarding Open AI on a weekly basis. I you know I hate to see it and don't
[17:48] you know I hate to see it and don't
[17:48] you know I hate to see it and don't necessarily want to make a joke about it
[17:49] necessarily want to make a joke about it
[17:49] necessarily want to make a joke about it but maybe I need to create a segment
[17:52] but maybe I need to create a segment
[17:52] but maybe I need to create a segment specifically for open AI uh you know
[17:56] specifically for open AI uh you know
[17:56] specifically for open AI uh you know safety issues and whistleblowers coming
[17:58] safety issues and whistleblowers coming
[17:58] safety issues and whistleblowers coming out because I believe we had a very
[18:01] out because I believe we had a very
[18:01] out because I believe we had a very similar story last week um Robinson or
[18:05] similar story last week um Robinson or
[18:06] similar story last week um Robinson or one of the guys that left chatpt came
[18:08] one of the guys that left chatpt came
[18:08] one of the guys that left chatpt came out you know in a similar fashion um to
[18:12] out you know in a similar fashion um to
[18:12] out you know in a similar fashion um to to provide some context here what these
[18:14] to provide some context here what these
[18:14] to provide some context here what these folks are talking about is um we now
[18:18] folks are talking about is um we now
[18:18] folks are talking about is um we now have and and have had for several years
[18:21] have and and have had for several years
[18:21] have and and have had for several years reasoning models. And so what these
[18:23] reasoning models. And so what these
[18:23] reasoning models. And so what these models do and you you've noticed it, but
[18:25] models do and you you've noticed it, but
[18:25] models do and you you've noticed it, but um when you ask a question, you can see
[18:27] um when you ask a question, you can see
[18:27] um when you ask a question, you can see it. Hey, it's it's going through a
[18:31] it. Hey, it's it's going through a
[18:31] it. Hey, it's it's going through a series of steps to find you the best
[18:33] series of steps to find you the best
[18:33] series of steps to find you the best answer. So, it's going to say, "Hey,
[18:35] answer. So, it's going to say, "Hey,
[18:35] answer. So, it's going to say, "Hey, analyzing the question, uh, searching
[18:38] analyzing the question, uh, searching
[18:38] analyzing the question, uh, searching the web, checking with this." Did it,
[18:41] the web, checking with this." Did it,
[18:41] the web, checking with this." Did it, you know, sometimes if you can if you
[18:42] you know, sometimes if you can if you
[18:42] you know, sometimes if you can if you click the drop down, you can see what
[18:44] click the drop down, you can see what
[18:44] click the drop down, you can see what it's uh saying, and it's like it it's
[18:47] it's uh saying, and it's like it it's
[18:47] it's uh saying, and it's like it it's kind of interesting to follow along.
[18:49] kind of interesting to follow along.
[18:49] kind of interesting to follow along. That being said, um, that uh little
[18:53] That being said, um, that uh little
[18:53] That being said, um, that uh little animation that you see in the thoughts,
[18:55] animation that you see in the thoughts,
[18:55] animation that you see in the thoughts, the quote unquote thoughts that the the
[18:57] the quote unquote thoughts that the the
[18:57] the quote unquote thoughts that the the model has, that's not the really that's
[19:00] model has, that's not the really that's
[19:00] model has, that's not the really that's not the the the trace logs. That's not
[19:02] not the the the trace logs. That's not
[19:02] not the the the trace logs. That's not the uh the actual thoughts. That's just
[19:05] the uh the actual thoughts. That's just
[19:05] the uh the actual thoughts. That's just a representation of what the model is
[19:08] a representation of what the model is
[19:08] a representation of what the model is doing at the time when it's trying to
[19:10] doing at the time when it's trying to
[19:10] doing at the time when it's trying to find the answer to your question. And so
[19:13] find the answer to your question. And so
[19:13] find the answer to your question. And so um what what these former open AI safety
[19:16] um what what these former open AI safety
[19:16] um what what these former open AI safety folks are saying is um we need to have
[19:20] folks are saying is um we need to have
[19:20] folks are saying is um we need to have the ability to actually see the logs and
[19:22] the ability to actually see the logs and
[19:22] the ability to actually see the logs and not necessarily the everyday person. The
[19:24] not necessarily the everyday person. The
[19:24] not necessarily the everyday person. The problem is and as far as I know
[19:27] problem is and as far as I know
[19:27] problem is and as far as I know internally
[19:28] internally
[19:28] internally uh at OpenAI definitely other uh
[19:32] uh at OpenAI definitely other uh
[19:32] uh at OpenAI definitely other uh researchers at at other labs or or other
[19:35] researchers at at other labs or or other
[19:35] researchers at at other labs or or other advanced AI researchers don't have
[19:37] advanced AI researchers don't have
[19:37] advanced AI researchers don't have access to u the actual you know logs
[19:41] access to u the actual you know logs
[19:41] access to u the actual you know logs that are going on that the actual chain
[19:43] that are going on that the actual chain
[19:43] that are going on that the actual chain of thought to be able to track and see
[19:45] of thought to be able to track and see
[19:45] of thought to be able to track and see the pattern uh or or see the um uh see
[19:49] the pattern uh or or see the um uh see
[19:49] the pattern uh or or see the um uh see the path that the model took to come up
[19:51] the path that the model took to come up
[19:51] the path that the model took to come up with an answer and how it reasoned. I
[19:54] with an answer and how it reasoned. I
[19:54] with an answer and how it reasoned. I mean, simple as that. And and you can
[19:56] mean, simple as that. And and you can
[19:56] mean, simple as that. And and you can imagine that's a problem if you don't
[19:58] imagine that's a problem if you don't
[19:58] imagine that's a problem if you don't know how the the model is thinking or
[20:01] know how the the model is thinking or
[20:01] know how the the model is thinking or why it's thinking a certain way. Um,
[20:04] why it's thinking a certain way. Um,
[20:04] why it's thinking a certain way. Um, then it makes it harder to protect
[20:07] then it makes it harder to protect
[20:07] then it makes it harder to protect against adversarial events. Um, and it
[20:10] against adversarial events. Um, and it
[20:10] against adversarial events. Um, and it sounds like from from this report that
[20:12] sounds like from from this report that
[20:12] sounds like from from this report that it's only when you really push the
[20:14] it's only when you really push the
[20:14] it's only when you really push the model. This isn't going to happen in a
[20:17] model. This isn't going to happen in a
[20:17] model. This isn't going to happen in a uh, you know, consumerf facing
[20:19] uh, you know, consumerf facing
[20:19] uh, you know, consumerf facing application. But still, it's it's it's
[20:21] application. But still, it's it's it's
[20:22] application. But still, it's it's it's worth bringing up. again like continued
[20:24] worth bringing up. again like continued
[20:24] worth bringing up. again like continued guard rails and this we've been hearing
[20:26] guard rails and this we've been hearing
[20:26] guard rails and this we've been hearing calls for this for a while like hey we
[20:28] calls for this for a while like hey we
[20:28] calls for this for a while like hey we should be able to look in and understand
[20:30] should be able to look in and understand
[20:30] should be able to look in and understand exactly what the the agent did um just
[20:33] exactly what the the agent did um just
[20:34] exactly what the the agent did um just in case something wrong happens the the
[20:36] in case something wrong happens the the
[20:36] in case something wrong happens the the risk in that is open AI exposes their
[20:39] risk in that is open AI exposes their
[20:39] risk in that is open AI exposes their secret sauce essentially you know they
[20:40] secret sauce essentially you know they
[20:40] secret sauce essentially you know they they if if they open that up to the
[20:42] they if if they open that up to the
[20:42] they if if they open that up to the public or they open that up to other
[20:44] public or they open that up to other
[20:44] public or they open that up to other researchers that's their unique I IP
[20:47] researchers that's their unique I IP
[20:47] researchers that's their unique I IP that's the only reason they're valuable
[20:49] that's the only reason they're valuable
[20:50] that's the only reason they're valuable right now uh and so I understand the
[20:51] right now uh and so I understand the
[20:51] right now uh and so I understand the risk from their standpoint standpoint as
[20:52] risk from their standpoint standpoint as
[20:52] risk from their standpoint standpoint as well. Um hopefully we continue to see
[20:55] well. Um hopefully we continue to see
[20:55] well. Um hopefully we continue to see developments here from a safety
[20:57] developments here from a safety
[20:57] developments here from a safety standpoint and and um push that forward.
[21:00] standpoint and and um push that forward.
[21:00] standpoint and and um push that forward. I would love for next week not to hear
[21:03] I would love for next week not to hear
[21:03] I would love for next week not to hear another whistleblower or not to hear
[21:05] another whistleblower or not to hear
[21:05] another whistleblower or not to hear about another safety incident um but
[21:07] about another safety incident um but
[21:07] about another safety incident um but maybe new protocols coming out of OpenAI
[21:10] maybe new protocols coming out of OpenAI
[21:10] maybe new protocols coming out of OpenAI that guards against us. I would love to
[21:11] that guards against us. I would love to
[21:11] that guards against us. I would love to see kind of a proactive approach. We'll
[21:13] see kind of a proactive approach. We'll
[21:13] see kind of a proactive approach. We'll see if they get there. I'm sure they
[21:15] see if they get there. I'm sure they
[21:15] see if they get there. I'm sure they they're doing a lot of work on on this
[21:17] they're doing a lot of work on on this
[21:17] they're doing a lot of work on on this front. Um, I guess, you know, to to put
[21:21] front. Um, I guess, you know, to to put
[21:21] front. Um, I guess, you know, to to put it to the the audience, the Austin
[21:23] it to the the audience, the Austin
[21:23] it to the the audience, the Austin Wilson show, uh, loyal followers. Um, do
[21:27] Wilson show, uh, loyal followers. Um, do
[21:27] Wilson show, uh, loyal followers. Um, do you think that labs should be forced to
[21:29] you think that labs should be forced to
[21:29] you think that labs should be forced to keep AI reasoning readable even if the
[21:32] keep AI reasoning readable even if the
[21:32] keep AI reasoning readable even if the that that costs on capability or maybe
[21:35] that that costs on capability or maybe
[21:35] that that costs on capability or maybe hurts their business? Be uh, curious to
[21:38] hurts their business? Be uh, curious to
[21:38] hurts their business? Be uh, curious to hear what you guys have to say.
[21:41] hear what you guys have to say.
[21:41] hear what you guys have to say. Microsoft shipped a Windows 11 sandbox
[21:44] Microsoft shipped a Windows 11 sandbox
[21:44] Microsoft shipped a Windows 11 sandbox to keep AI agents on a leash and open
[21:47] to keep AI agents on a leash and open
[21:47] to keep AI agents on a leash and open pre-orders on their new Nvidia powered
[21:51] pre-orders on their new Nvidia powered
[21:51] pre-orders on their new Nvidia powered Surface AI PCs. Uh at its October 7th
[21:55] Surface AI PCs. Uh at its October 7th
[21:55] Surface AI PCs. Uh at its October 7th Windows and Surface event, Microsoft
[21:57] Windows and Surface event, Microsoft
[21:57] Windows and Surface event, Microsoft made uh Microsoft execution containers
[22:01] made uh Microsoft execution containers
[22:01] made uh Microsoft execution containers generally available on Windows 11. MXC
[22:05] generally available on Windows 11. MXC
[22:05] generally available on Windows 11. MXC is an operating system enforced
[22:07] is an operating system enforced
[22:07] is an operating system enforced isolation where developers declare what
[22:09] isolation where developers declare what
[22:09] isolation where developers declare what an agent can touch. This is referred to
[22:11] an agent can touch. This is referred to
[22:11] an agent can touch. This is referred to as a sandbox. Uh Microsoft also opened
[22:15] as a sandbox. Uh Microsoft also opened
[22:15] as a sandbox. Uh Microsoft also opened pre-orders for uh their Nvidia powered
[22:19] pre-orders for uh their Nvidia powered
[22:19] pre-orders for uh their Nvidia powered uh RTX Spark Surface Laptop Ultra uh
[22:24] uh RTX Spark Surface Laptop Ultra uh
[22:24] uh RTX Spark Surface Laptop Ultra uh from $2,600 with up up to 128 GB of
[22:28] from $2,600 with up up to 128 GB of
[22:28] from $2,600 with up up to 128 GB of unified memory uh shipping October 16th.
[22:31] unified memory uh shipping October 16th.
[22:31] unified memory uh shipping October 16th. a Surface RTX Spark Dev Box, uh, which
[22:35] a Surface RTX Spark Dev Box, uh, which
[22:35] a Surface RTX Spark Dev Box, uh, which would be the more advanced version would
[22:37] would be the more advanced version would
[22:37] would be the more advanced version would be about $6,000,
[22:40] be about $6,000,
[22:40] be about $6,000, uh, ready to go in November, just in
[22:42] uh, ready to go in November, just in
[22:42] uh, ready to go in November, just in time for Christmas. Are we going to see
[22:45] time for Christmas. Are we going to see
[22:45] time for Christmas. Are we going to see some, uh, Windows, Nvidia
[22:49] some, uh, Windows, Nvidia
[22:49] some, uh, Windows, Nvidia powered GPU, smart computers under the
[22:53] powered GPU, smart computers under the
[22:53] powered GPU, smart computers under the Christmas tree this year? Maybe. Um,
[22:55] Christmas tree this year? Maybe. Um,
[22:56] Christmas tree this year? Maybe. Um, Windows 11 gets a built-in AI agent
[22:58] Windows 11 gets a built-in AI agent
[22:58] Windows 11 gets a built-in AI agent sandbox and um, Nvidia chip uh, Surface
[23:02] sandbox and um, Nvidia chip uh, Surface
[23:02] sandbox and um, Nvidia chip uh, Surface AI PCs go on pre-order. Agents uh,
[23:06] AI PCs go on pre-order. Agents uh,
[23:06] AI PCs go on pre-order. Agents uh, running loose on PCs is kind of the next
[23:09] running loose on PCs is kind of the next
[23:09] running loose on PCs is kind of the next security headache that we want to look
[23:10] security headache that we want to look
[23:10] security headache that we want to look out for. Microsoft is putting
[23:12] out for. Microsoft is putting
[23:12] out for. Microsoft is putting containment into the operating system
[23:14] containment into the operating system
[23:14] containment into the operating system while betting on local models running on
[23:17] while betting on local models running on
[23:17] while betting on local models running on uh, Nvidia Silicon. Microsoft pitches
[23:19] uh, Nvidia Silicon. Microsoft pitches
[23:19] uh, Nvidia Silicon. Microsoft pitches the Surface devices for running large AI
[23:22] the Surface devices for running large AI
[23:22] the Surface devices for running large AI models locally and says MXC lets Windows
[23:26] models locally and says MXC lets Windows
[23:26] models locally and says MXC lets Windows uh not the agent decide what files and
[23:29] uh not the agent decide what files and
[23:29] uh not the agent decide what files and network it can reach. Uh pre-orders are
[23:32] network it can reach. Uh pre-orders are
[23:32] network it can reach. Uh pre-orders are live. Get to your local Windows retailer
[23:35] live. Get to your local Windows retailer
[23:36] live. Get to your local Windows retailer and go buy some Surface Pros with Nvidia
[23:40] and go buy some Surface Pros with Nvidia
[23:40] and go buy some Surface Pros with Nvidia GPUs. Um the laptop was announced
[23:44] GPUs. Um the laptop was announced
[23:44] GPUs. Um the laptop was announced earlier this year, May 31st. So, the
[23:46] earlier this year, May 31st. So, the
[23:46] earlier this year, May 31st. So, the news um is really just about price
[23:49] news um is really just about price
[23:49] news um is really just about price availability and some more of the
[23:50] availability and some more of the
[23:50] availability and some more of the details. Nvidia's speed claims are its
[23:53] details. Nvidia's speed claims are its
[23:54] details. Nvidia's speed claims are its own um and the sandbox only helps if if
[23:57] own um and the sandbox only helps if if
[23:57] own um and the sandbox only helps if if agent makers are using them. So, uh it's
[24:00] agent makers are using them. So, uh it's
[24:00] agent makers are using them. So, uh it's not enforced uh necessarily on the
[24:03] not enforced uh necessarily on the
[24:03] not enforced uh necessarily on the average user. This product is more
[24:06] average user. This product is more
[24:06] average user. This product is more focused and and um on developers and um
[24:11] focused and and um on developers and um
[24:11] focused and and um on developers and um uh you know people testing out local
[24:13] uh you know people testing out local
[24:13] uh you know people testing out local models. So this is when we talk about um
[24:16] models. So this is when we talk about um
[24:16] models. So this is when we talk about um Mistral or we talk about Beam or these
[24:18] Mistral or we talk about Beam or these
[24:18] Mistral or we talk about Beam or these other uh open-source models. Um this is
[24:23] other uh open-source models. Um this is
[24:23] other uh open-source models. Um this is exactly the use case for for um this new
[24:26] exactly the use case for for um this new
[24:26] exactly the use case for for um this new Windows Surface Pro. And actually at 128
[24:29] Windows Surface Pro. And actually at 128
[24:29] Windows Surface Pro. And actually at 128 GB of RAM uh a very uh um a very
[24:35] GB of RAM uh a very uh um a very
[24:35] GB of RAM uh a very uh um a very reasonable price in the kind of $2,000
[24:38] reasonable price in the kind of $2,000
[24:38] reasonable price in the kind of $2,000 range uh for a machine like that, which
[24:41] range uh for a machine like that, which
[24:41] range uh for a machine like that, which is awesome. um for the everyday user,
[24:45] is awesome. um for the everyday user,
[24:45] is awesome. um for the everyday user, but you probably would not use this
[24:47] but you probably would not use this
[24:47] but you probably would not use this today. But that being said, what what
[24:49] today. But that being said, what what
[24:49] today. But that being said, what what we're seeing a lot is um in the future,
[24:53] we're seeing a lot is um in the future,
[24:53] we're seeing a lot is um in the future, there's probably going to be a world
[24:54] there's probably going to be a world
[24:54] there's probably going to be a world where every MacBook, every computer that
[24:58] where every MacBook, every computer that
[24:58] where every MacBook, every computer that you have, your your phone um they are
[25:01] you have, your your phone um they are
[25:01] you have, your your phone um they are going to have the capability to run
[25:05] going to have the capability to run
[25:05] going to have the capability to run models on these. And we've talked about
[25:07] models on these. And we've talked about
[25:07] models on these. And we've talked about it in this p in the past where um you
[25:10] it in this p in the past where um you
[25:10] it in this p in the past where um you know an AI model is comprised of now
[25:14] know an AI model is comprised of now
[25:14] know an AI model is comprised of now today upwards of trillions of
[25:16] today upwards of trillions of
[25:16] today upwards of trillions of parameters. When you ask a question the
[25:19] parameters. When you ask a question the
[25:20] parameters. When you ask a question the the machine needs to run through each of
[25:21] the machine needs to run through each of
[25:22] the machine needs to run through each of those parameters to bring back an
[25:24] those parameters to bring back an
[25:24] those parameters to bring back an answer. You got to imagine you have to
[25:25] answer. You got to imagine you have to
[25:25] answer. You got to imagine you have to have a powerful machine to to run that.
[25:28] have a powerful machine to to run that.
[25:28] have a powerful machine to to run that. Um what we're seeing is these are
[25:30] Um what we're seeing is these are
[25:30] Um what we're seeing is these are getting smaller and smaller. Naturally
[25:32] getting smaller and smaller. Naturally
[25:32] getting smaller and smaller. Naturally this, you know, anybody uh not new to
[25:36] this, you know, anybody uh not new to
[25:36] this, you know, anybody uh not new to technology knows that the form factor
[25:38] technology knows that the form factor
[25:38] technology knows that the form factor gets smaller and smaller, the
[25:39] gets smaller and smaller, the
[25:39] gets smaller and smaller, the capabilities get better and better.
[25:41] capabilities get better and better.
[25:41] capabilities get better and better. That's why when we talk about the new
[25:42] That's why when we talk about the new
[25:42] That's why when we talk about the new Haiku model, Anthropic getting cheaper
[25:45] Haiku model, Anthropic getting cheaper
[25:45] Haiku model, Anthropic getting cheaper and faster is really cool because uh we
[25:48] and faster is really cool because uh we
[25:48] and faster is really cool because uh we know that those smaller models are
[25:50] know that those smaller models are
[25:50] know that those smaller models are getting smarter and smarter and are
[25:52] getting smarter and smarter and are
[25:52] getting smarter and smarter and are would be capable to run on the local
[25:54] would be capable to run on the local
[25:54] would be capable to run on the local machine. Why would you want to run that
[25:56] machine. Why would you want to run that
[25:56] machine. Why would you want to run that locally? Well, that's a great question.
[25:58] locally? Well, that's a great question.
[25:58] locally? Well, that's a great question. in the future. Um, and maybe not too
[26:01] in the future. Um, and maybe not too
[26:01] in the future. Um, and maybe not too distant future, right? Uh, there's a lot
[26:03] distant future, right? Uh, there's a lot
[26:03] distant future, right? Uh, there's a lot of sensitive stuff that you don't want
[26:05] of sensitive stuff that you don't want
[26:05] of sensitive stuff that you don't want to and and today hopefully, right,
[26:07] to and and today hopefully, right,
[26:07] to and and today hopefully, right, you're not typing that stuff into chat
[26:10] you're not typing that stuff into chat
[26:10] you're not typing that stuff into chat GPT or Claude. Um, you're not uploading
[26:14] GPT or Claude. Um, you're not uploading
[26:14] GPT or Claude. Um, you're not uploading um, you know, personal information.
[26:17] um, you know, personal information.
[26:17] um, you know, personal information. Um,
[26:18] Um,
[26:18] Um, you want to keep a lot of this safe,
[26:20] you want to keep a lot of this safe,
[26:20] you want to keep a lot of this safe, especially in, you know, uh, from a
[26:23] especially in, you know, uh, from a
[26:23] especially in, you know, uh, from a security standpoint. This is this is
[26:25] security standpoint. This is this is
[26:25] security standpoint. This is this is important to do. And so what we'll see
[26:27] important to do. And so what we'll see
[26:27] important to do. And so what we'll see um both from a security standpoint but
[26:29] um both from a security standpoint but
[26:29] um both from a security standpoint but also from um just a a usability and and
[26:32] also from um just a a usability and and
[26:32] also from um just a a usability and and functionality standpoint um running the
[26:35] functionality standpoint um running the
[26:35] functionality standpoint um running the intelligence on a local machine. You
[26:37] intelligence on a local machine. You
[26:37] intelligence on a local machine. You know imagine you're uh a small business
[26:40] know imagine you're uh a small business
[26:40] know imagine you're uh a small business owner and and you've got some
[26:42] owner and and you've got some
[26:42] owner and and you've got some proprietary processes and practices. You
[26:45] proprietary processes and practices. You
[26:45] proprietary processes and practices. You don't necessarily want to be putting
[26:47] don't necessarily want to be putting
[26:47] don't necessarily want to be putting that into um a you know um online uh AI
[26:53] that into um a you know um online uh AI
[26:53] that into um a you know um online uh AI model, right? um being able to keep that
[26:56] model, right? um being able to keep that
[26:56] model, right? um being able to keep that on your local computer. Um keeping that
[26:59] on your local computer. Um keeping that
[26:59] on your local computer. Um keeping that data specific to you, you know, think
[27:02] data specific to you, you know, think
[27:02] data specific to you, you know, think about healthcare data or financial data
[27:04] about healthcare data or financial data
[27:04] about healthcare data or financial data or anything like that. Being able to
[27:06] or anything like that. Being able to
[27:06] or anything like that. Being able to keep that personal is going to be really
[27:07] keep that personal is going to be really
[27:07] keep that personal is going to be really important. We'll still have web- based
[27:10] important. We'll still have web- based
[27:10] important. We'll still have web- based models um that will probably communicate
[27:12] models um that will probably communicate
[27:12] models um that will probably communicate back and forth, but you've got to
[27:14] back and forth, but you've got to
[27:14] back and forth, but you've got to imagine if you have chat GPT powered
[27:17] imagine if you have chat GPT powered
[27:17] imagine if you have chat GPT powered like everything and it doesn't need an
[27:19] like everything and it doesn't need an
[27:19] like everything and it doesn't need an internet connection, you can use it
[27:21] internet connection, you can use it
[27:21] internet connection, you can use it wherever you want. uh there's a lot a
[27:23] wherever you want. uh there's a lot a
[27:23] wherever you want. uh there's a lot a lot of u of interesting things that you
[27:25] lot of u of interesting things that you
[27:26] lot of u of interesting things that you can do there and um and we're seeing
[27:27] can do there and um and we're seeing
[27:28] can do there and um and we're seeing kind of the progression here right as
[27:29] kind of the progression here right as
[27:29] kind of the progression here right as models get faster smarter smaller
[27:33] models get faster smarter smaller
[27:33] models get faster smarter smaller cheaper um you know great that open AI
[27:36] cheaper um you know great that open AI
[27:36] cheaper um you know great that open AI is solving these you know crazy math
[27:40] is solving these you know crazy math
[27:40] is solving these you know crazy math problems and doing really advanced
[27:42] problems and doing really advanced
[27:42] problems and doing really advanced things but the everyday person does not
[27:44] things but the everyday person does not
[27:44] things but the everyday person does not need that right um and so on on the
[27:47] need that right um and so on on the
[27:47] need that right um and so on on the other end of the spectrum we're getting
[27:49] other end of the spectrum we're getting
[27:49] other end of the spectrum we're getting more capable models um closer to us
[27:52] more capable models um closer to us
[27:52] more capable models um closer to us instead of sending that away to a data
[27:55] instead of sending that away to a data
[27:55] instead of sending that away to a data center um you know somewhere
[27:58] center um you know somewhere
[27:58] center um you know somewhere uh across the country to power our uh AI
[28:02] uh across the country to power our uh AI
[28:02] uh across the country to power our uh AI uses. Um curious is anybody looking at
[28:07] uses. Um curious is anybody looking at
[28:07] uses. Um curious is anybody looking at buying um one of these new surfaces? Um
[28:10] buying um one of these new surfaces? Um
[28:10] buying um one of these new surfaces? Um would love to get feedback if you've
[28:12] would love to get feedback if you've
[28:12] would love to get feedback if you've seen one or tested one or anything like
[28:14] seen one or tested one or anything like
[28:14] seen one or tested one or anything like that. Uh would be really interested to
[28:17] that. Uh would be really interested to
[28:17] that. Uh would be really interested to see what's happening there. Uh finally,
[28:19] see what's happening there. Uh finally,
[28:19] see what's happening there. Uh finally, Google just opened its synth ID checker
[28:22] Google just opened its synth ID checker
[28:22] Google just opened its synth ID checker to the public. Uh so anyone can test
[28:24] to the public. Uh so anyone can test
[28:24] to the public. Uh so anyone can test whether an image, video or song came
[28:26] whether an image, video or song came
[28:26] whether an image, video or song came from AI. Uh Google launched synth
[28:30] from AI. Uh Google launched synth
[28:30] from AI. Uh Google launched synth id.com. It's a public site where anyone
[28:32] id.com. It's a public site where anyone
[28:32] id.com. It's a public site where anyone can upload an image, video or audio file
[28:35] can upload an image, video or audio file
[28:35] can upload an image, video or audio file to check for synth ID. Um which is its
[28:38] to check for synth ID. Um which is its
[28:38] to check for synth ID. Um which is its AI watermark. Since Google io their
[28:42] AI watermark. Since Google io their
[28:42] AI watermark. Since Google io their event in 2025, the tool had been limited
[28:46] event in 2025, the tool had been limited
[28:46] event in 2025, the tool had been limited to just journalists and researchers. Uh,
[28:48] to just journalists and researchers. Uh,
[28:48] to just journalists and researchers. Uh, synth ID marks um output from Google's
[28:51] synth ID marks um output from Google's
[28:51] synth ID marks um output from Google's nano banana, vio and lia models and
[28:55] nano banana, vio and lia models and
[28:55] nano banana, vio and lia models and openai, Nvidia and cacao also support
[28:59] openai, Nvidia and cacao also support
[28:59] openai, Nvidia and cacao also support it. Uh, Google says it handles about 1
[29:02] it. Uh, Google says it handles about 1
[29:02] it. Uh, Google says it handles about 1 million verification requests per day.
[29:05] million verification requests per day.
[29:05] million verification requests per day. Uh, that's not a small number. Google's
[29:07] Uh, that's not a small number. Google's
[29:07] Uh, that's not a small number. Google's AI watermark director uh excuse me
[29:10] AI watermark director uh excuse me
[29:10] AI watermark director uh excuse me detector synth ID is now open to anyone
[29:14] detector synth ID is now open to anyone
[29:14] detector synth ID is now open to anyone at synth ID.com uh with midterms close
[29:18] at synth ID.com uh with midterms close
[29:18] at synth ID.com uh with midterms close and AI fakes everywhere a free public
[29:21] and AI fakes everywhere a free public
[29:21] and AI fakes everywhere a free public way to check media is a real tool but
[29:24] way to check media is a real tool but
[29:24] way to check media is a real tool but only for content that carries the
[29:26] only for content that carries the
[29:26] only for content that carries the watermark. Uh Google says anyone can now
[29:29] watermark. Uh Google says anyone can now
[29:29] watermark. Uh Google says anyone can now verify whether media was made with AI
[29:31] verify whether media was made with AI
[29:31] verify whether media was made with AI tools uh that embed synth ID. The site
[29:35] tools uh that embed synth ID. The site
[29:35] tools uh that embed synth ID. The site is live and supports common image,
[29:38] is live and supports common image,
[29:38] is live and supports common image, video, and audio formats. Google reports
[29:40] video, and audio formats. Google reports
[29:40] video, and audio formats. Google reports about a million verifications. Um, it
[29:43] about a million verifications. Um, it
[29:43] about a million verifications. Um, it only detects the synth ID. Microsoft and
[29:45] only detects the synth ID. Microsoft and
[29:45] only detects the synth ID. Microsoft and Meta use their own standards, and open
[29:47] Meta use their own standards, and open
[29:47] Meta use their own standards, and open source or stripped content won't show
[29:49] source or stripped content won't show
[29:49] source or stripped content won't show up. So, a quote unquote clean result
[29:52] up. So, a quote unquote clean result
[29:52] up. So, a quote unquote clean result doesn't prove necessarily that a human
[29:56] doesn't prove necessarily that a human
[29:56] doesn't prove necessarily that a human uh made it. Uh this is we were just
[29:59] uh made it. Uh this is we were just
[29:59] uh made it. Uh this is we were just talking about this the other day but um
[30:03] talking about this the other day but um
[30:03] talking about this the other day but um uh great that Google opened this up to
[30:06] uh great that Google opened this up to
[30:06] uh great that Google opened this up to anybody for anybody to check. These are
[30:08] anybody for anybody to check. These are
[30:08] anybody for anybody to check. These are the type of protocols and this is when
[30:10] the type of protocols and this is when
[30:10] the type of protocols and this is when um you open source technology like this
[30:13] um you open source technology like this
[30:13] um you open source technology like this you see um kind of the the power of the
[30:17] you see um kind of the the power of the
[30:17] you see um kind of the the power of the community and the improvement of the
[30:19] community and the improvement of the
[30:19] community and the improvement of the technology overall. Right? So um uh with
[30:23] technology overall. Right? So um uh with
[30:24] technology overall. Right? So um uh with Google open sourcing this there's a good
[30:26] Google open sourcing this there's a good
[30:26] Google open sourcing this there's a good um or or making this generally available
[30:29] um or or making this generally available
[30:29] um or or making this generally available there's a a good chance that this
[30:31] there's a a good chance that this
[30:31] there's a a good chance that this becomes the standard protocol. So any AI
[30:33] becomes the standard protocol. So any AI
[30:33] becomes the standard protocol. So any AI image moving forward would have this
[30:35] image moving forward would have this
[30:36] image moving forward would have this synth ID image any video any audio uh or
[30:39] synth ID image any video any audio uh or
[30:39] synth ID image any video any audio uh or or music um would have this synth ID. So
[30:42] or music um would have this synth ID. So
[30:42] or music um would have this synth ID. So you you know not detectable to the human
[30:45] you you know not detectable to the human
[30:45] you you know not detectable to the human ear or the human eye but uh when you run
[30:48] ear or the human eye but uh when you run
[30:48] ear or the human eye but uh when you run it through a machine um the the um the
[30:52] it through a machine um the the um the
[30:52] it through a machine um the the um the other models would be able to detect
[30:54] other models would be able to detect
[30:54] other models would be able to detect that it's not uh human created that it
[30:57] that it's not uh human created that it
[30:57] that it's not uh human created that it is AI created. Uh again this offers more
[31:00] is AI created. Uh again this offers more
[31:00] is AI created. Uh again this offers more protections. Um uh this this hearkens
[31:03] protections. Um uh this this hearkens
[31:03] protections. Um uh this this hearkens back to like chat GBT coming through
[31:05] back to like chat GBT coming through
[31:05] back to like chat GBT coming through with their text watermark like which is
[31:08] with their text watermark like which is
[31:08] with their text watermark like which is uh great to see. Um, I I feel like there
[31:10] uh great to see. Um, I I feel like there
[31:10] uh great to see. Um, I I feel like there would be much more danger in the the
[31:12] would be much more danger in the the
[31:12] would be much more danger in the the audio and the video uh aspect of this.
[31:15] audio and the video uh aspect of this.
[31:15] audio and the video uh aspect of this. Uh, what's interesting is, yeah, for
[31:18] Uh, what's interesting is, yeah, for
[31:18] Uh, what's interesting is, yeah, for generally speaking, if if I'm sitting
[31:19] generally speaking, if if I'm sitting
[31:19] generally speaking, if if I'm sitting back and going, "Okay, well, so then I'm
[31:21] back and going, "Okay, well, so then I'm
[31:21] back and going, "Okay, well, so then I'm going to have to check every single
[31:22] going to have to check every single
[31:22] going to have to check every single video uh or, you know, screen every
[31:26] video uh or, you know, screen every
[31:26] video uh or, you know, screen every phone call to make sure that it's really
[31:28] phone call to make sure that it's really
[31:28] phone call to make sure that it's really my grandma calling asking for $100." uh
[31:32] my grandma calling asking for $100." uh
[31:32] my grandma calling asking for $100." uh um uh know what what would happen in the
[31:36] um uh know what what would happen in the
[31:36] um uh know what what would happen in the future is this would be programmatic and
[31:38] future is this would be programmatic and
[31:38] future is this would be programmatic and and would be automated and we're already
[31:40] and would be automated and we're already
[31:40] and would be automated and we're already seeing that in a lot of social media
[31:42] seeing that in a lot of social media
[31:42] seeing that in a lot of social media where um you see tags uh that that say
[31:46] where um you see tags uh that that say
[31:46] where um you see tags uh that that say AI generated some of that is uh manually
[31:49] AI generated some of that is uh manually
[31:49] AI generated some of that is uh manually done but but you can imagine a future
[31:50] done but but you can imagine a future
[31:50] done but but you can imagine a future where uh that's just automated any AI
[31:53] where uh that's just automated any AI
[31:54] where uh that's just automated any AI generated uh content is uh is tagged um
[31:58] generated uh content is uh is tagged um
[31:58] generated uh content is uh is tagged um and and again going back to local models
[32:01] and and again going back to local models
[32:01] and and again going back to local models you know having app running on your
[32:02] you know having app running on your
[32:02] you know having app running on your phone on your local computer is great.
[32:05] phone on your local computer is great.
[32:05] phone on your local computer is great. Uh just improves safety, improves trust
[32:07] Uh just improves safety, improves trust
[32:07] Uh just improves safety, improves trust overall, right? Like in this uh this age
[32:10] overall, right? Like in this uh this age
[32:10] overall, right? Like in this uh this age of AI slop um being able to trust the
[32:13] of AI slop um being able to trust the
[32:14] of AI slop um being able to trust the content that we see and just to know,
[32:15] content that we see and just to know,
[32:16] content that we see and just to know, okay, hey, like I'm fine with AI
[32:17] okay, hey, like I'm fine with AI
[32:18] okay, hey, like I'm fine with AI content, right? But just to know um and
[32:20] content, right? But just to know um and
[32:20] content, right? But just to know um and and not have to sit back and go, wait,
[32:24] and not have to sit back and go, wait,
[32:24] and not have to sit back and go, wait, is that AI? You know, like um you know,
[32:27] is that AI? You know, like um you know,
[32:27] is that AI? You know, like um you know, not not getting a text from your
[32:30] not not getting a text from your
[32:30] not not getting a text from your parents. um
[32:33] parents. um
[32:33] parents. um you know with some ridiculous AI video
[32:35] you know with some ridiculous AI video
[32:35] you know with some ridiculous AI video asking if it's uh if it's real. I saw
[32:38] asking if it's uh if it's real. I saw
[32:38] asking if it's uh if it's real. I saw one the other day about it was a crazy
[32:40] one the other day about it was a crazy
[32:40] one the other day about it was a crazy one where it was like a uh a car wreck
[32:43] one where it was like a uh a car wreck
[32:43] one where it was like a uh a car wreck between a Coca-Cola truck and a Mentos
[32:45] between a Coca-Cola truck and a Mentos
[32:45] between a Coca-Cola truck and a Mentos truck and it just like you know blew
[32:48] truck and it just like you know blew
[32:48] truck and it just like you know blew blew up all over the place and it's like
[32:50] blew up all over the place and it's like
[32:50] blew up all over the place and it's like your parents texting to to ask if uh you
[32:54] your parents texting to to ask if uh you
[32:54] your parents texting to to ask if uh you know if you saw this latest crazy video.
[32:57] know if you saw this latest crazy video.
[32:57] know if you saw this latest crazy video. Anyway, so um great to see this out of
[32:59] Anyway, so um great to see this out of
[32:59] Anyway, so um great to see this out of Google. uh appreciate them making this
[33:02] Google. uh appreciate them making this
[33:02] Google. uh appreciate them making this generally available. Um you know,
[33:05] generally available. Um you know,
[33:05] generally available. Um you know, wondering if you guys would actually run
[33:07] wondering if you guys would actually run
[33:07] wondering if you guys would actually run this yourself um or if you would just
[33:10] this yourself um or if you would just
[33:10] this yourself um or if you would just wait until this is kind of automated in
[33:13] wait until this is kind of automated in
[33:13] wait until this is kind of automated in there. Uh feel free to share your
[33:16] there. Uh feel free to share your
[33:16] there. Uh feel free to share your thoughts with us. Man, we covered a lot
[33:18] thoughts with us. Man, we covered a lot
[33:18] thoughts with us. Man, we covered a lot today. It was again a really fun day.
[33:21] today. It was again a really fun day.
[33:21] today. It was again a really fun day. Um, I I really love to see the the
[33:25] Um, I I really love to see the the
[33:25] Um, I I really love to see the the through lines through the through the
[33:26] through lines through the through the
[33:26] through lines through the through the stories that are coming out. Um, and and
[33:28] stories that are coming out. Um, and and
[33:28] stories that are coming out. Um, and and what we're talking about on a day-to-day
[33:30] what we're talking about on a day-to-day
[33:30] what we're talking about on a day-to-day basis and then hopefully uh you guys are
[33:33] basis and then hopefully uh you guys are
[33:33] basis and then hopefully uh you guys are are um uh gaining quite a bit from just
[33:37] are um uh gaining quite a bit from just
[33:37] are um uh gaining quite a bit from just being able to share the context of why
[33:39] being able to share the context of why
[33:39] being able to share the context of why these uh stories matter uh why they
[33:42] these uh stories matter uh why they
[33:42] these uh stories matter uh why they matter to us as individuals in our
[33:44] matter to us as individuals in our
[33:44] matter to us as individuals in our day-to-day and where this leads to in
[33:46] day-to-day and where this leads to in
[33:46] day-to-day and where this leads to in the future. uh and then just to grasp a
[33:49] the future. uh and then just to grasp a
[33:49] the future. uh and then just to grasp a little bit of understanding um on how
[33:53] little bit of understanding um on how
[33:53] little bit of understanding um on how all of this works and how it works
[33:55] all of this works and how it works
[33:55] all of this works and how it works together and just get a little bit
[33:57] together and just get a little bit
[33:57] together and just get a little bit better uh dayto-day. That's that's my
[33:59] better uh dayto-day. That's that's my
[33:59] better uh dayto-day. That's that's my goal. That's what I'm doing here
[34:01] goal. That's what I'm doing here
[34:01] goal. That's what I'm doing here hopefully and and and hopefully adding
[34:03] hopefully and and and hopefully adding
[34:03] hopefully and and and hopefully adding some value to you guys and you're
[34:04] some value to you guys and you're
[34:04] some value to you guys and you're enjoying it and we're having fun. We're
[34:06] enjoying it and we're having fun. We're
[34:06] enjoying it and we're having fun. We're having fun. This is great. I love it.
[34:08] having fun. This is great. I love it.
[34:08] having fun. This is great. I love it. Um, come back here tomorrow, 6:00 a.m.
[34:12] Um, come back here tomorrow, 6:00 a.m.
[34:12] Um, come back here tomorrow, 6:00 a.m. We're doing AI news, real content, and
[34:16] We're doing AI news, real content, and
[34:16] We're doing AI news, real content, and I'm your human in the loop, Austin
[34:18] I'm your human in the loop, Austin
[34:18] I'm your human in the loop, Austin Wilson. Let's go. Have a great day.
[35:04] Hey, hey, hey. Hey. Hey. Hey.
[37:47] Heat. Heat. N.
[38:08] Heat. Heat.
[38:57] Heat. Heat. N.
[39:38] Heat. Heat.
[39:44] Most
[40:04] people
[40:07] believe
[40:10] I don't appreciate what he's doing
[40:12] I don't appreciate what he's doing because I've seen
[40:19] See
[40:32] this is correct.
[40:36] Yeah,
[40:39] Yeah, this can't be right.
[40:42] this can't be right.
[40:42] this can't be right. 2us 182
[40:47] consider
[40:54] what did you think of the possibility?
[41:04] So uh so I said the primary
[41:08] So uh so I said the primary on your own
[41:13] shows up
[41:20] I can't
[41:29] open%.
[41:52] together.
[42:06] units.
[42:21] My kids now
[42:24] My kids now have to go
[42:26] have to go
[42:26] have to go to
[42:38] We're both pretty good.
[42:49] No one seems to get fired. Absolutely
[42:52] No one seems to get fired. Absolutely ridiculous.
[43:04] 199
[43:23] is a three
[43:24] is a three So
[43:32] some
[43:53] every
[43:59] I'll just keep in mind
[44:49] What you're Thank you.
[45:40] Show is done um and live on X. But for
[45:45] Show is done um and live on X. But for some reason there was an issue with
[45:48] some reason there was an issue with
[45:48] some reason there was an issue with um
[45:50] um
[45:50] um going live on YouTube. So I'll just post
[45:52] going live on YouTube. So I'll just post
[45:52] going live on YouTube. So I'll just post the YouTube video direct there.