AI Impersonates Humans So Well It Fooled 48% Of People

0:00We have a brand new update in Claude

0:02because we've just got our access to

0:04mods. Claude mods has dropped. And on

0:06top of that, really importantly, we have

0:08a milestone in AI video generation

0:11conversation where effectively an AI

0:13video model has officially passed the

0:15touring test, meaning that more people

0:18than fewer were a not able to identify

0:20that they were talking to an AI. Nick,

0:22shall we actually begin with Griffin

0:25Travis's new model? Talk about what this

0:27means. Uh, actually it was 48% of people

0:30who talked to it thought it was a real

0:32human. So roughly 50/50 split.

0:34>> Pretty insane to think about. I mean, I

0:36think we all knew that we were heading

0:38in this direction, but to see it happen

0:40so soon, bonkers. Um, I personally

0:42maintain that Vanessa here, quote

0:45unquote, still has a couple of tells. So

0:48for me, it's not, you know, total cat

0:51out of the hat so far. I mean, I can see

0:52she's putting up her fingers and stuff,

0:54but there's something weird with her

0:55eyes. Her smile looks a little bit off.

0:57Still, I could absolutely see this fool

0:59uh some granny's you know so I mean if

1:02you guys don't already have some sort of

1:03like safe password setup with most of

1:05your family would definitely implement

1:07that. Now my big question for this stuff

1:09and I don't mean this in a bad way to

1:11Tavis because you know I think that this

1:13sort of development is impressive is

1:15what are the positive use cases of a

1:19technology like this?

1:23>> That's a good question. That's a really

1:24good question. I first of all I think

1:26the first thing to do is sign your alarm

1:27bells. Tell every grandmother within

1:30five miles of Texas beware. Get ready.

1:34>> Just Texas.

1:35>> Yeah, just Texas specifically Texas.

1:37It's cool. Who I think is worth looking

1:40at here is I was playing around with the

1:41previous version of Tapers cuz this is

1:44out right now in early access.

1:46>> I actually and I actually played with

1:48the previous version and the previous

1:50version they they basically can't move

1:52their hands. they can't move their body

1:54and that was one of his biggest limiting

1:56factors, Nick, is that it was kind of

1:57like but here she can like lift her

2:00thumb lift her thumb up and stuff. So

2:02you might even see it here for example

2:03on this uh demo maybe he gets a bit like

2:05that's got like an AI feel to it right

2:08mainly because he's not justiculating

2:10the use cases I mean honestly I suppose

2:14it's probably just consumer apps right

2:16like just having like a individual that

2:19you can talk to to guide you through

2:20things that there's learning

2:22applications I would imagine

2:25>> but

2:28yeah yeah I think yeah that's what I was

2:30trying to get at just like I mean I

2:31could see learning applications in a

2:32situation like this where you have like

2:34some 3D CGI tutor or something you know

2:36sure whatever I could see some

2:38applications where I don't know people

2:39are or rather the intelligence is

2:41stylized and it's clear that like while

2:43I anthropomorphized it's not like a

2:44human being but the ones where like

2:46they're meant to look just like human

2:48beings. I mean what is the comparative

2:50value there? I I don't know man. I

2:52really don't know what the comparative

2:53actual value of that is aside from just

2:55like literally and I hate to say it just

2:57talking mostly to old people that

2:58wouldn't be able to tell the difference.

3:00There are probably some legitimate use

3:01cases where you can talk to old people

3:03that wouldn't be able to tell the

3:04difference. You know, maybe people in

3:05care homes like um you know that company

3:07they were working with. I don't know,

3:08maybe you know a 3D CGI happy fun smiley

3:12face robot thing would freak them the

3:14[ __ ] out. So, you want it to seem like

3:15there's a human being on the other end

3:16of the line. But still, it all feels

3:18like it's based on deception to me. And

3:19I think that's the part that's sort of

3:21unsettling. Um all of this is in some

3:23way, shape, or form deceiving. No. Well,

3:25when you said that, I just thought to

3:27myself, well, what are all the times

3:29when I'm on a screen like this with this

3:30square? And I was like, well, we could

3:32get the one on the podcast.

3:35We could get an AI Jack and AI Nick to

3:37join us. So, guys, if it ever happens

3:39that one was not here, this could never

3:41happen again. We can tav this Nick in.

3:44We can

3:44>> blink twice if she needs help.

3:47>> Two of them and like give them and like

3:48make one like more funny than the other

3:50one. I'd like to make make them very

3:52disagreeable. Um, so we we could you can

3:55real time have AI podcast like people

3:57podcast. The other thing I'm thinking as

3:59well is like wouldn't it be really funny

4:00if you have like you know like a call in

4:03one of our communities and then we're

4:04like just bring an AI in who just like

4:08knows where all the stuff like I know

4:09just like has like loads of knowledge or

4:11something or

4:13is just funny or I don't know you could

4:15bring him into calls or something like

4:17or even with clients.

4:19It could it could also be a liability

4:20though if he starts glitching out

4:21because hey it's no reflection on the

4:23skill of what we bring to the table. But

4:25I I think there's some cool thing I

4:26think it will replace note takers

4:28eventually like can you imagine having

4:29like AIS on the meeting

4:32>> but I feel like if we get to that if we

4:34get to that point though Jack would you

4:35even want to do the meeting? I mean

4:36you're already at the point I'm pretty

4:38sure where like half your eating

4:39meetings you're just like leaving and

4:40you're like okay that could have been an

4:41email so [ __ ] let's do a call when

4:43you actually have something valuable

4:44enough to jump on a call for me to share

4:46with. Like

4:47>> don't you think this is one of those

4:48really low bandwidth situations that

4:49like and that's why I'm thinking like

4:51sure kids Yeah. And then really old

4:53people both of which have like

4:55legitimate intellectual bottlenecks

4:58around their ability to communicate.

4:59Like sure I can get that but like for

5:01like anybody else I don't know maybe I'm

5:04just crazy here. If you guys see

5:07>> Yeah,

5:08>> I know what you mean. I know what you

5:09mean. I think like it's probably got a

5:11few things like Yeah. I mean it's it's

5:14an interesting one. I think it's very

5:15cool. Like, if it genuinely is

5:17indistinguishable,

5:20then that's really [ __ ] amazing.

5:21Like, that's really impressive. Like,

5:23that that's I'm sure we'll find all the

5:25cool use cases for it. But, I mean, how

5:27would you feel, Nick, if someone booked

5:29in a 15-minute call, then they sent

5:30their like tapis avatar?

5:33>> A feeling that'd be such a burn

5:36>> for you to do that to somebody and like

5:38I have like a consultation link or

5:40something and it still costs like a,000

5:42bucks. Somebody pays it sends the tapest

5:44thing.

5:46>> I do that to somebody.

5:48>> Yeah. The digital avatar of Nick. Hello

5:51there.

5:52>> I'm Well, the the real the real Nick is

5:54out there working with his clients. I'm

5:56here to help you really understand.

5:58>> We're too bit too busy to make it out of

6:00the call.

6:02>> I can't make it. I might have a tall

6:04joke. He knows what he's talking about.

6:07>> Dude, I I actually built an AI Jack and

6:09Sun. Dude, I actually preferred AI Jacks

6:11over Jack's response. I was like, I

6:13don't know. It's like both a burn and a

6:15compliment simultaneously, right? That

6:17it's great they like it, but also like

6:19they preferred AI Jack comment. So, I

6:21would probably iron unironically think

6:23it'd be funny, Nick, to send it to

6:25meetings on.

6:27>> That is funny. Jack, can I derail this

6:29totally and talk about something that we

6:30were not planning on talking about at

6:31all?

6:32>> I I think we need to better talk to me.

6:34So earlier today, the Pope tweeted, and

6:37I think he tweeted basically because of

6:40me.

6:41>> Whoa. So this is the this is the

6:44situation. You guys know how a few days

6:46ago I made this ex post that was um

6:49well, they're calling it the the worst

6:51ratio of all time.

6:53>> They're not, are they?

6:55>> I don't think they are calling it the

6:56worst. I don't think a single human

6:58being has been ratioed as I've been

7:00ratioed.

7:02got [ __ ] barely 3,000 likes something

7:03with 14 million. Anyway, um immediately

7:06after, you know, a bunch of other people

7:08started and then guess what Pope Leo I

7:1014th said, "In this era of artificial

7:12intelligence, it is becoming urgent to

7:14distinguish human art from what machines

7:15produce. There's an onlogical

7:17difference, even before an aesthetic

7:18one, between art and what a machine can

7:20generate through statistical calculation

7:21based on millions of images created by

7:23others. Algorithms lack the spark of

7:25humanity. For this reason, the church

7:27wants to renew an alliance with artists

7:28and cultural institutions to safeguard

7:30our humanity. Jack, please tell me I'm

7:32not crazy here.

7:33>> So, so it looks like Nick has caused the

7:36Pope to essentially um renew Hang on,

7:40guys. He's caused the Pope to create an

7:42alliance with artists and cultural

7:44institutions to safeguard humanity due

7:46to the evil forces that Nick is

7:48perpetrating essentially impropagating

7:50with his with his heresy.

7:52Um,

7:54>> dude, is that is that yours or his that

7:56he's

7:57>> I made this one. I made this one. I

7:59thought cuz come on. Like where who else

8:00is this going to be about? Dude, I feel

8:02like I lit the match.

8:04>> You're beep you're beeping with the pope

8:05now, bro. This is outrageous.

8:07>> Do you know what my friend sent me? He's

8:08like, "There was only one man that I

8:10ever could have expected that could

8:12bring the lefties and the Holy Catholic

8:15Empire together, and it was you."

8:18I think every every political group,

8:21religion, and even world leader have

8:24united in hatred against NV.

8:28>> It's it's it's it's literally ushered in

8:31an era of of world peace guys because

8:34everyone hates it that much.

8:36>> Freaking hilarious, dude. I mean, like,

8:38dude, that's

8:40>> that's pretty cool. I mean, I used to

8:41live next to the Pope. If it was the

8:42same Pope, I could have asked him like I

8:43could have

8:44>> Did you really? Oh, like in Rome.

8:47>> Yeah.

8:47>> Yeah. That's funny. You know, no, nobody

8:49on the pod knows about like your Roman

8:51adventure. Very few people know that

8:52like you legitimately lived in Italy for

8:54years,

8:55>> dude. It was Yeah, I did. For a year,

8:56dude. It was um Papa Franchesco at the

8:58time. Very, very cool, man. Used to walk

9:00through there every day on the way to

9:01uni. It was It was cool, man. I think he

9:03would have um he would have lectured me,

9:05man. He would have thought that'd be

9:06great.

9:07>> Yeah, Franchesco, man. I I miss I miss

9:10him.

9:11>> Bless his heart. Nick, should we talk

9:13about Jev? Um, we got to talk about the

9:15Pope as well later on. And we're also

9:16guys giving away a Mac Mini or Cash

9:19equivalent today. Um, so that's going to

9:21be happening in a few minutes. We're

9:22going to talk about what's going on with

9:24that. But then Jev is back for more Jev

9:27related things. Should we talk about

9:28Clefft and what is fresh with that?

9:31>> Let's talk about Que.

9:33>> Sorry

9:34targeting Jev. Dude, I'm going to get

9:36cancel. It's coming out Cle. So Clefft

9:39is Cloudfare's free to download decision

9:41model. picks amongst fixed options

9:43returns probabilities just like Jev. Uh

9:46Jev is basically the hosted version of

9:47it. You can grab it through open router

9:49and type safe that website. So before

9:51this month cloud had no release models

9:53in it category. It now offers cleft and

9:56the smaller cleft flash. I think flash

9:58has become the standard thing in the

9:59industry now. It's hosted model servers

10:01using the name software connection

10:03format as Jev. So builders can download

10:06both model files under a permissive

10:07commercial license. But Cloud Pair's

10:10five benchmarks win apparently, Nick, it

10:12is sending Jev to the bleachers

10:15allegedly.

10:16>> Yeah, I mean the main, you know,

10:17headline here is just how fast KFFT

10:19flashes compared to Jev. Keep in mind

10:22Jev was already touted for being

10:24extraordinarily responsive and just

10:26orders of magnitude better than like

10:28traditional large language models at

10:29routing and decision-m and that was at

10:31500 milliseconds. Cleft flash, which has

10:33come out less than 2 and 1/2 weeks after

10:35the initial Jev drop, is at 38.8 8

10:38milliseconds which is about 115th of Jev

10:42which was 115th of large language

10:44models. I mean the growth here is

10:45absolutely insane. So lots of

10:47optimization pressures going on. You can

10:49feasibly have you know a router based

10:52model like in your computer right now

10:54looking over every single place that

10:56your mouse moves on the screen and then

10:58predicting what apps you want to open

11:00simply based off of that. Um people are

11:03building these sorts of applications.

11:04You can have real-time painting of stuff

11:07using, you know, just voice

11:08transcription, real-time app

11:10development, simply using your voice.

11:12You can do eye tracking with this stuff.

11:13I mean, I mean, the possibilities are

11:14actually genuinely insane. So, I think

11:17we're going to see a Cambrian explosion

11:19of cool apps in the next couple of

11:20weeks.

11:21>> They are. And guys, this isn't a little

11:23nudge and a little nudge here, n that.

11:24This is a quantum I actually I didn't

11:26say quantum lip. I'm getting carried

11:27away. This is a crazy increase. I mean,

11:31Nick, we're talking at least 10 times

11:32faster. at least a 10x multiple in

11:34speed, probably 15x actually, which is

11:36crazy. So that's wild. One of the cool

11:39use cases people have done dynamic ad

11:41blockers, right? Like if you want to

11:43remove all the LinkedIn AI generated

11:45content, you can do that. LinkedIn ad

11:48swap buster

11:49>> deletes the entire

11:50>> basically a white becomes a white page

11:53that you scroll through basically.

11:55>> But but it's cool because when you can

11:57make hyper fast decisions four 500 times

11:59faster, actually 2,000 times faster than

12:01Astro, right? but is significantly

12:03cheaper. It unlocks new economic

12:06opportunities. So very very cool that

12:08we're getting this innovation in this

12:09kind of second class of model nomic.

12:12>> Yeah, it's nice. It's nice to see like

12:13more diversity. Um because I think a lot

12:15of people have tried to shove large

12:18language models which are basically

12:19round pegs into kind of circular holes

12:23if if you catch my drift. Like we've

12:25used them for virtually everything. They

12:26really are the hammer that makes

12:27everything else look like a nail. Um we

12:29use them for all sorts of dynamic

12:31routing. We use them for like our voice

12:32agents. We use them for text generation

12:35editing. But the reality is like you can

12:37train custom slightly different

12:39architecture models that are just way

12:41way way better. And that's really just

12:43you know what this was. This was an

12:44inroads to you know a world where

12:47artificial intelligence is soon going to

12:48be embedded in literally everything. You

12:50could have, you know, cleft flash or a

12:52derivative of it in your light and there

12:55could be a little camera connected to

12:56your light connected to cleft flash such

12:58that all you need to do is literally

12:59like look over at your light and your

13:01light would know that you want to turn

13:02it off. I mean, like the possibilities

13:04here are virtually endless. So, all of

13:07this model development could stop today

13:08and we'd still be weaving this into the

13:10architecture around us for the next

13:11thousand years.

13:13>> It's so it's so freaking cool. the the

13:15question you have to ask yourself cuz if

13:16you're thinking like okay I I like these

13:18use cases but what do I do the question

13:20to ask yourself is is there a task

13:22you're doing number one and number two

13:24can you map out all of the possible

13:28decisions in the decision because if you

13:30can it can be cleft it can be jev

13:33because je if you say look Jeb here only

13:35I it could only be 15 things I know that

13:38for a fact

13:39>> then Jev can be your main guy or cleft

13:41in this instance so

13:43>> it's cool Nick it's cool I think

13:44probably even though like we're really

13:46set and we're really we're kind of

13:48nerding out and I think this is still in

13:49nerd territory. This is not in consumer

13:52application like it's not like you know

13:54your barber Jason's not going to be like

13:55hey did you check out cleft they dropped

13:57yesterday like it's not quite like

13:59mainstream yet but this is a very very

14:02exciting update.

14:03>> Yeah. Yeah. I mean I feel like this is

14:05one of those things that's just like it

14:06it can't talk so it's like naturally

14:08going to be less interesting but it is

14:09extremely powerful.

14:11Yeah. I mean it makes you wonder what

14:13else if there's a third category like

14:15what the next one would look like.

14:17>> That's pretty interesting. Um

14:20I mean like

14:22every everything so far is like based

14:23off next token prediction, right?

14:26Um well sorry I guess like actually

14:29there are multiple different types of

14:30models. So the models for images right

14:32now are called diffusion based models,

14:34right? So diffusion is like a

14:36fundamentally different process than the

14:37process that is occurring with large

14:38language models right now. And um it's

14:41basically where like you take an image

14:43and then you kind of like convert it

14:45into noise. So it's like a really noisy

14:47background and then you train the model

14:48on the difference between the image and

14:50then the noise and then you can reverse

14:52that and then generate any image just

14:53out of noise in a prompt. Really cool

14:55technology. Um but that like that's like

14:58the image models. They're using that for

14:59music too. So like Suno is using

15:01diffusion which is kind of cool. Um but

15:04like aside from that it's just large

15:05language models now this new decision

15:07router model. Like like is there

15:08anything else? No, I mean like because

15:10Jev is just a different output modality,

15:13right? It's like instead of text, you're

15:15getting binary basically and then you're

15:18right diffusion image. What is left? I

15:20mean, it's going to see we're going to

15:21see, dude. We're going to see what the

15:22hell's going to go on, but these are

15:24pretty exceptional updates. Um Nick, we

15:26also have an exceptional update for one

15:28member of the Stack Podcast. Should we

15:30Should we run the Rumble? Should we run

15:32the Suma? Guys, if you don't know, um,

15:35on the Slack podcast, pretty much the

15:37entirety of last month, every week, we

15:39gave away a Mac Mini or Cash equivalent

15:41because that's just how we do things

15:43here. So, we've got a link, we've got a

15:45giant sumo ring that you're all going to

15:47compete to fight on, and my last man or

15:49woman standing is going to win this Mac

15:50Mini. Some say some said we were mad

15:52men. They said we were crazy to do this,

15:56>> but I said we had to do it.

15:57>> I'm I'm one of I'm one of the they think

16:00this is insane.

16:02Let's start the drawing. We have 1,277

16:05entries from exactly 500 people. So

16:08median was about two and a half entries

16:10per person. Let's start the draw.

16:13>> So what's interesting on this is that

16:15your expected value is over there's like

16:17basically like a dollar per comment.

16:21>> Oh yeah, I didn't think about that.

16:22That's true.

16:23>> You guys make a dollar every time you

16:25comment.

16:25>> You basically make a dollar for

16:26commenting. Yeah.

16:27>> Yeah.

16:28>> Make a dollar for commenting. Yeah.

16:29>> It's like free money. The eliminations

16:31are are wild right now. So Nick, you

16:33you're wearing quite a cool uh colorful

16:35shirt, Nick. Do you share with the

16:36people what what is it? A special shirt

16:38today.

16:39>> Yeah, it's a special shirt. My uh

16:41>> what's the design?

16:42>> Mom bought my father and I matching

16:44shirts for her birthday. We're in

16:46Hawaii,

16:47>> so we uh went and we celebrated and holy

16:50[ __ ] dude. Talk about the fallen.

16:53>> God damn. Okay, final 17. Here we go.

16:56>> Then who we got? Oh my gosh,

16:58>> that was f That was a fast one. Ryman

17:00Y2K

17:01>> flew right by. Yeah, we got a Y2K lover,

17:04Ryman. Um, nice job, Ryman.

17:08Congratulations.

17:09>> I kind of want to like

17:10>> Congrat

17:12Have we ever watched this on 8X, Jack?

17:15>> Oh, it's fast.

17:16>> If I run this again, is somebody else

17:18going to win?

17:20>> We need to give them something though if

17:21they did.

17:22>> Yeah, that's true. So, I guess we can.

17:23Well, Ryman uh-Y2K,

17:26congratulations, my friend. my uh

17:30compatriate in order to claim your Mac

17:31Mini. If you could please update that

17:33very account, Ryman-Y2K

17:36such that we can see your email address

17:38on the account, Jack or I will reach out

17:41to you in the next 7 days and pay you.

17:43And we do have to do it this way, as

17:45unfortunate as it is, because we've

17:47we've actually in the last week,

17:48something Jack, I haven't told you, is

17:50we've had a lot of people trying to scam

17:51the stacked podcast.

17:53>> Oh my god.

17:54>> By pretending that they're one of the

17:55winners. And so we can't have that. So,

17:57you do need to update your actual

17:59profile. There are there is a lot of

18:01evil in the world. That's what Jack and

18:02I were busy fighting today. That's why

18:04we were 12 hours behind. Um, and so,

18:07yeah, just make sure that you actually

18:08update your email. Congratulations. We

18:11don't have a video from last week's

18:12winner, but we certainly will next week.

18:15Uh, how you feeling about this whole

18:16thing, Jack?

18:18>> I think it's good, man. I think it's

18:19crazy. We had the evil forces were

18:21trying to take stack podcast and we had

18:23to stop him with the with the bring in

18:25the Pope to get it done. Well, we got it

18:27done. We got it done in the end.

18:29>> Um I you know I I wonder like guys tell

18:31it down below in the comment this video

18:33like what it is that you do like I would

18:35love to know who's actually watching

18:37just like maybe what you do where you

18:39were. I just curious to know have we got

18:42doctors, engineers? Have we got like um

18:46cracked out 15year-olds who were trying

18:48to like build trillion dollar companies

18:51or do we have evil forces Nick? We have

18:53evil forces in our comment section

18:54trying to take us down, trying to take

18:55the Stack Podcast Max down there.

18:57>> So, do this way for a reason. And

19:00>> we're we're going to figure out what's

19:01going on there.

19:02>> Yeah, true, Jack. I couldn't help but

19:04notice there was that episode that you

19:05did solo and it got 10,000 views. And

19:08then there's the episode that I did solo

19:09and it got 12,000 views. What do we

19:13>> Dude, there must be about that.

19:15>> The algorithm must be broken, Nick. I

19:16don't know what's going on with that.

19:17>> Must be a broken algo. Yeah,

19:19>> we got we got we got to zoom out and

19:20look at all the solo episodes. I get the

19:22proper data.

19:23>> Yeah, for sure.

19:24>> Nick, that was the sour spot. Sonet 4.5

19:27was so wholesome indeed. I didn't know

19:30um so many people felt the same. I'd

19:32rather Sonic have stayed like that

19:33trying to compete with that person on

19:34coding haiku and speed. Nick, you want

19:37to bring people in with your thoughts on

19:38this friendly neighborhood chart badge?

19:40>> Yeah, it's unfortunate, but basically

19:42they're retiring sauna 4.5. Um and it's

19:45similar what happened when OpenAI

19:47retired GPT40. A lot of people had been

19:49using 4.5. They really liked 4.5 more as

19:52like a relationship sort of friend, you

19:54know, somebody to listen to the somebody

19:56with personality than a predictive

19:58coding partner. When Anthropic announced

20:00that they were going to retire it, a lot

20:02of people got angry and it looks like

20:03Sleeping Troubles is is one of them. So,

20:06it's a it's a wider conversation around

20:07how there are a lot of people using

20:09models not for economically valuable

20:10ends or at least not chiefly for

20:12economically valuable ends that keep

20:13using the old ones simply because they

20:15like how they sound, how they talk to

20:16them. M

20:17>> and when you retire one, it's kind of

20:18like, you know, putting taking old

20:20yeller out the back, putting him down.

20:23>> Yeah. And I'm not saying Sleeping

20:24Troubles is in a relationship with Sonic

20:274.5, but like you get these strong

20:30people do like they get like like either

20:33a AI buddy or like this is AI

20:36girlfriends, Nick, that have been built

20:37on models and they have a very

20:38particular communication style and then

20:40they got like dropped up and then it's

20:42like the whole world like completely

20:43changed. People People do build

20:45relationships and affinities with these

20:48models 100%.

20:49>> Agreed. Yeah. Grocky Grock Grock. Hey

20:53Claude, am I the Mac winner? Nope. It's

20:55shoulder day on your left.

20:58White House action. Okay, we're getting

21:00a lot of comments here. Ladies and

21:01gentlemen, this fella clearly wants the

21:03Mac Mini. Um, Maris hearing Fable 5.5

21:06rumblings. Jack, have you heard of

21:08anything about Fable 5.5?

21:10>> I have not heard rumblings on Fable 5.5

21:13Pro, actually. No, I mean, are we just

21:14skipping out 5.234 these days? Like,

21:17what the hell is going on there? The

21:18heck is happening?

21:20>> Yeah, it's uh weird. Everybody's saying

21:21like, "Whoa, my Clawude Fable outputs

21:24have upgraded overnight. This must be

21:26Fable 5.5 and it's all over X." And I

21:29don't know if it's a big nothing burger.

21:32>> It's an unconfirmed rumored AI model

21:35social media users are claiming is being

21:37secretly tested and rooted behind the

21:39scenes by Anthropic.

21:42So they're think people are basically

21:43saying Nick that there's stealth rooting

21:45going on. Users in X and re Reddit claim

21:47Anthropic is silently rooting some user

21:49interfaces from 5.1 to a hidden 5.5

21:53fable back end.

21:55>> Yeah. Yeah. It's sneaky, you know.

21:57Allegedly.

21:59>> Allegedly.

22:00>> Allegedly.

22:03>> Dude, I commented just last week you

22:05needed to stop your sny your snarky

22:08sarcasm dissing Google. And now here we

22:11go. You're glazing Gemini. The hypocrisy

22:13is disappointing.

22:14>> Damn. It's almost like Gemini dropped a

22:17new model or something.

22:18>> It's brutal.

22:20>> I'd say guys, our opinions are subject

22:22to change based on information

22:24>> and they and they uh we don't we don't

22:26dislike any big provider per se, but we

22:29we make some moves and we have we have

22:30racy takes and we make magic happen. So

22:33guys, big huge congratulations to the

22:34winner. Have a beautiful and blessed day

22:36guys and we will catch you in the next

22:37episode. See you later.