Subscribe
Share
Share
Embed
This week on the show, BJ is joined by David Gerard of Pivot to AI. Come listen about how the AI bubble, like crypto, is a scam funded by the venture capital bros, and how that scam works. If we're all eating out of soup cans this time next year, you may hear in this episode why. Learn more about our show at Stupid Sexy Privacy.com.
Creators and Guests
Host
Host
Rosie Tran
Rosie Tran is one of the fastest rising stars in the entertainment business! Originally from New Orleans, Louisiana, she moved to Hollywood to pursue her career as a professional entertainer. The stand up comedian, writer, podcast personality, and actress has toured internationally, at comedy clubs, colleges, and overseas for the USO in Europe and the Middle East.
Editor
Andrew
I am the Editor of all things on the Stupid Sexy Privacy Podcast.
What is Stupid Sexy Privacy?
Stupid Sexy Privacy is a miniseries about how to protect yourself from fascists and weirdos. Your host is comedian Rosie Tran, and the show is written by information privacy expert B.J. Mendelson. Every episode is sponsored by our friends at DuckDuckGo. Tune in every Thursday night βor Friday morning if you're nasty β at 12 am EST to catch the next episode.
BJ (00:00)
All right. David, would you like to take a moment just to introduce yourself to the listeners of Stupid Sexy Privacy?
David Gerard (00:07)
Hello, I'm David Gerrard. I write pivot to AI.com and mostly the YouTube channel attached to it, Pivot to AI. And it's also a podcast version. And basically it's a daily thing on whatever stupid nonsense the AI bubble has come up with. I used to write about cryptocurrency, but basically that died in the ass around 2024 and we got really bored and decided to do AI instead.
Because it turned out the crypto bros were pivoting to AI. So we decided to do the same and it was a good name for it.
BJ (00:43)
That's right.
Yeah, let me ask you real quick about that. Cause we, I used to be a crypto guy. Like I, in the early 2010s liked the idea of it, liked the concept behind it. But then I found very quickly it became hijacked by what we know, what you call now crypto bros or the tech bros and the bro-ogargs or whatever you want to call them. β And it's just become a giant Ponzi scheme for the wealthy. And I was curious if that's sort of where you've settled on your opinion of it.
David Gerard (01:17)
on what? opinion on bitcoins.
BJ (01:20)
just the crypto in general.
David Gerard (01:22)
All right. No, I mean, it's like, it sort of started that way. It was started by libertarian tech bros who encapsulate the California ideology, β which we now have the most rabid version of. But on the other hand, I was actually surprised and pleased to discover in 2025, 2026, a lot of those people are actually libertarians. And like, for instance, they
BJ (01:50)
Alright.
David Gerard (01:52)
feel deep concern about what Trump is doing, that sort of thing, and they're outraged by it, and their cryptos don't help them, but it turns out that they are very sincerely liberty minded, and that's much better than the alternative. I can argue a lot of things with libertarianism, but the basic idea, it's not, it has a lot of sympathy.
BJ (01:58)
That's right.
Yes, no.
Yeah, know, for me, β as left as, I wouldn't quite say I'm in the anarchist category, but I'm pretty left, right? Pretty liberal. And for me, just knowing at least American history with discrimination from the banks β towards anyone that wasn't white or anyone that wasn't wealthy, you know, that to me is what of what attracted me to the idea of crypto as, we can stop doing this. We could stop redlining. And of course, it just did not pan out.
Adieu.
David Gerard (02:46)
Yeah, they came up with
the right questions and all their answers were on crack.
BJ (02:50)
So that's right. I love what you're doing with Pivot.ai. I found that I'm now a devotee of it. Since we, you I heard and reached out to you a while ago and then life kept getting in the way and then finally we've had this chance to speak. What, I'm curious about the mechanics of what you're doing, because what you're doing is this very bold and like you're doing daily coverage of something that is very complex for a lot of people.
β And you're doing it in such a way where it's described so well and described with the sarcasm that it deserves. And I'm just a little curious about the behind the scenes for producing Pivot AI. Like what led to you saying, okay, we're gonna get in front of a camera and Monday through Friday, we're just gonna do this.
David Gerard (03:38)
Right. So Amy Castor, my co-writer and I, we started Pivot to AI because we got bored by crypto. We've been covering it for a few years, working together. We write well together. And so we thought, let's do an AI site. The whole idea was Web3 is going great, but it's AI. And Molly White went, cool, maybe people will stop asking me. And so the idea was we did something every day. It was seven days a week at first.
BJ (03:56)
Okay.
to
David Gerard (04:07)
of short thing, no more than 250 words say anything over that. gets a read more and just do that every day. And you don't run out of material that way. You can just do a short piece on something. This is when we were doing crypto stuff. was like 2000 words every few days. So it was a lot easier. And around the end of 2024, I got made redundant. So I'm now
BJ (04:25)
Right.
David Gerard (04:36)
YouTube bum, YouTube influenza. And I had this for Christmas 2024, my family gave me a ring light and said, stop ranting at us, rant on TikTok. So ring lights are incompatible with wearing glasses though, as you get little googly eyes projected on your lenses and it's most amusing. But the basic idea stuck in my head and I realized I could turn these little stories into
BJ (04:49)
Right.
Yes.
David Gerard (05:05)
minutes of video every day. So we end up with our present format. And the way I do it is I have a thing I want to cover and I'll often go on my socials and say, hey, what's pissing you off today, my fellow AI hater bros? And they'll tell me what the hot news is. Sometimes it's obvious, like someone throws a Molotov at Sam Altman's house. I've then got to cover that next. And that turns into an extensive dive into the rationalist subculture and
BJ (05:26)
Yes.
David Gerard (05:34)
what they actually believe, all of which is nuts. And so I'm faced with having to explain really quite complicated things in a simple, accessible mainstream format. And that's quite good because that's literally what I was doing with the Bitcoin stuff. It's also like Amy does a few things for us occasionally still. She basically went off to be retired. She hates the word retired because it makes it sound like she's not doing anything. But
BJ (06:04)
sure.
David Gerard (06:04)
more or less. So β I'm dragging your into stuff occasionally. since then it's been just me, but it's basically, I find out the idea. sort of spend all day messing around, not getting any work done. Then I in a frantic burst of energy from about 4pm onwards, I write it, record it and then produce it. And you know, writing is easy. Doing video is really long winded.
BJ (06:34)
Yes.
David Gerard (06:34)
About
an hour, a minute or so. yeah. And the YouTube has taken off like a rocket and I'm very pleased. I actual money. It's a shock. And I also do the YouTube as a podcast because, and there's still the blog posts and I tend to write that once I've done the YouTube and they're like three separate audiences. They're like completely disjoint. People who like one hate the others and don't understand how anyone could like that.
BJ (07:00)
Right, yeah.
That's an interesting thing that we've encountered because we have our sister site, the Monroe and Gazette, and Stupid Sexy Privacy. And what we found is that people that like the articles don't like the podcast, and people that like the podcast don't like the video. And so we found, even though it's the same content, there's definitely, people have their preferences as to where they want their information. So something I'm really curious about,
David Gerard (07:20)
β yeah.
BJ (07:32)
You are basically writing about a moving target, because as you mentioned, something will come up like Sam Altman having a Molotov cocktail thrown at his house. β There was the whole thing. I think I had sent you β with Hegseth and Anthropic. There was that bubbling news story that came to the surface. Do you find that there's anything that you want to write about, you just haven't been able to crack it yet?
David Gerard (07:58)
Um, I have like a huge long single Google doc, which, um, I keep all my notes and it's like a hundred thousand words by now. And that's just the thing I use. And I often have notes on stuff like, think the longest winded was I, uh, Amy and I scribbled some notes on Ubeam in 2024, which was not AI at all. Right. It was, but it was another venture capital, uh, boondoggle.
BJ (08:07)
Right?
David Gerard (08:27)
from the mid 2010s where they told you, said you could recharge your phone using ultrasound. Every physicist listening to you just went, what? And this is not possible, but they sold it really hard to a whole bunch of venture capitalists. And I eventually put that one out during the Christmas break, Christmas break week as a nice extra. It's not AI, but it's totally AI because
BJ (08:34)
I remember.
David Gerard (08:57)
I view the AI bubble that we are in is fundamentally a venture capital scam. That's where it comes from. That's how it works. If it was market forces, I predicted, and Edzitron also predicted this same way, talking to people in AI, that all this would crash around the end of 2024. It didn't. I think we were wrong because we assumed market forces.
BJ (09:04)
Yes.
David Gerard (09:26)
This isn't market forces. This is a scam. This is a large complex fraud. there's my, one of my favorite books. It's one of these books. I tell everyone to buy lying for money by Dan Davies. There's a UK and US edition. Get both of them. They, cause they got different stories in him. He's, he's got a whole theory of frauds and it involves that.
BJ (09:34)
Yes.
David Gerard (09:53)
frauds always grow bigger and bigger and spiral out as they try to cover up fraud with more fraud and they grow until they can't. So that's what the AI bubble is doing. It's a sort of multi-company Enron where they are writing, this is the way it works. They all write large numbers on paper justified by private company equity.
BJ (10:09)
Yes. Yes.
David Gerard (10:22)
And they value that private equity in billions of dollars and write down a number with the dollar sign in front and nine digits. And it's, it's all fake. That's not actual money. It's not even actually properly valued. Like with stocks, they're sort of imaginary, but also there's an actual market where you can buy and sell this stuff. That doesn't apply to open AI shares. example, before it was like even weirder because it was the
rights to a percentage of open AI's future profit, if any ever happens. And they were selling that as being open AI shares and equity. And it wasn't really any of those things. It's, it's now equity. So it's a bit more real. And if they do a public offering, it will actually be stock. So you can talk about it as being much closer to being money, but the actual, the valuations are basically headline marketing lies.
They're all so you can write down a number with a lot of digits and a dollar sign in front. So you can borrow against it and that's your excuse. Or you can use leverage for other things and that's your excuse. Same with Wall Street Bitcoin. They have all this huge amounts of Bitcoin that no one will ever buy because there is not that much of a market in actual dollars for Bitcoins. We know there is not much of a market in actual dollars for Bitcoins because we can see Coinbase
puts in its SSE filings, its trading figures for Bitcoin. And they're going down, down, down slowly every quarter because people aren't into Bitcoins anymore. I don't know what they're doing. What's happened to the suckers of yesteryear? What has happened? There's a Rube and Fool gap that needs explaining. And I think they might have gotten into sports gambling or something, but it works very similarly, you know. But Wall Street Bitcoin, where they build huge Bitcoin
BJ (12:08)
Yes.
David Gerard (12:20)
company trusts and so on. That's the same sort of thing. You get this commodity that no one can possibly buy because the market literally doesn't exist. There aren't enough dollars to buy these things. But you say, these are worth a squillion dollars. I'm going to borrow against them. And then you do that. And it's a whole lot of paper shuffling and eventually all the bills come due, generally at once. And so AI works the same way.
In AI, the spiringly outwards is more and more and more venture capital being put into these money burning companies that they, they give you the imaginary value, but they burn actual dollars. Like they spend them, they send them to Nvidia and to power companies and turn them into e-waste and heat. And this spirals outwards until it can't anymore. And I predict
I suspect this will be around 2027 sometime. It'll all fall over. It should be sooner, but these guys must keep going because they have imaginary book values that they must justify. So they have a strong motivation to throw all the money they have at this thing happening. And that is basically the scam. A single example, there was one funding round where Softbank put in
20 odd billion dollars of actual money that they got from bankers and fundraising and so on. And they gave that to OpenAI and what they got from it was $43 billion of imaginary AI equity value that they wrote on their books. They literally exchanged real money for imaginary money and their stock price then went up. So Softbank stockholders thought this was a good move. So yeah, this is
Now you might think this is all just imaginary nonsense bollocks and you'd be 100 % correct because it is. But when, when it all crashes, the, you'll note the economy numbers, the numbers that look really good are basically held up by a few large companies swapping, kiting checks with each other more or less. And the actual economy where people live has a lot of bad numbers coming out of it, which like
Employment is slowing down, rent is through the roof, homelessness is rising hugely. Little indicators like car loan defaults are going through the roof. So the actual economy where you and I, the mere peons live is turbo screwed, but you know, the number is up for the moment. So it's not officially a recession. You've never had it so good.
BJ (15:05)
Let me ask you, because my fear is you can't separate the regime. I'm not supposed to say their actual name. β You can't separate the American regime from AI β right now. They're pretty heavily invested in it. The same way the family's pretty heavily invested in crypto. But I'm a little curious, because you're overseas. Do you have the same sort of fear that I have?
where this AI thing is going to burst and it's going to take down the American economy with it because all we have is AI gambling and cars, right? That's sort of our economy. Do you feel a little more insulated over where you are in the UK?
David Gerard (15:49)
So I'm in the UK and we've sort of been in that β stagnating stage for many years now, where some people are doing very well, but most of us are sort of not. β So it can get worse and it will get worse. I think that this, if the American economy goes out, then it's going to hit all the other economies. It's going to hit Europe and the UK, very much so.
There's stuff you can do about that. It depends whether they want to do it. So I think that it will be bad and we're headed for great depression too. I've had pushback on this from people where I say, yes, this is going to be great depression too. No, really the whole great depression too. think, well, there'll be a recession, then it'll recover. I think not because all the institutions that would bring us back from that are being white entered.
BJ (16:29)
Yes.
Yes.
David Gerard (16:48)
Jay Powell, this β was Trump's previous appointee for chair of the Federal Reserve. He's a central banker. He's generally held to have done his job competently. know, β the economy hasn't caught fire and exploded, so good. So Trump's pick to replace him is Kevin Warsh, who's a Bitcoiner. And worse than that, he's serious about the
Austrian economics, libertarian gold bug thing. He can, he can give you theory about it. So in great depression one countries came out of the great depression as they left the gold standard. So he's from the team who want to restore the gold standard. So he's precisely the wrong man in the wrong job at the wrong time. So we're headed to what I believe is called the
BJ (17:41)
Absolutely.
David Gerard (17:45)
Cool Zone.
BJ (17:47)
Is that what we're gonna call it? I mean, I totally agree with you. Yeah, I feel the same way in that. β
David Gerard (17:54)
It's all going to be sort of terrible. I don't have investment advice for people because this risks trashing the actual US dollar internationally as well. The US economy has 300 million people. So that's quite a lot of people and they're doing work and so on. But managing an economy is sort of working out how to distribute the little green ration tickets and keep everyone's doing stuff. So
BJ (17:56)
Yes.
David Gerard (18:24)
A recovery is possible, it's whether these guys are up for the job or... It depends when people start sending fire to things. Not that I'm advocating civil unrest. That would be rude and impolite and unpleasant. But it would be best if that doesn't even come up as a question.
BJ (18:31)
You
Yeah, β I'm there with you. I'm at the point where, and I'm always speaking for myself here, not how we're at. My concern is that Americans are very, β it's not that they're dumb, it's just that we're very insulated and we don't have a good understanding of the wider world around us in terms of even how our state functions, let alone the federal government. there's just a lot of people that.
David Gerard (18:47)
Don't go Fed posting on podcasts.
BJ (19:07)
you would think would be very smart and knowledgeable. They have no idea how badly the federal government has just been collapsed in the last year or so. And so I'm a little more optimistic that the EU and the UK will have smart people to ride this out. And I'm a little more concerned about, I think we kind of need this to be blunt. I kind of feel like America kind of needs a nice sharp kick in the ass.
David Gerard (19:36)
Yeah, it won't be a sharp kick in the ass. It'll be a long-term disabling, awful nightmare. β Nobody wants this except for idiots who don't know what they're doing. Unfortunately, they're in power right now. I think we can state as objective facts that Trump's administration team is not very good at many of the jobs they do.
BJ (19:40)
Yes. Yes.
Correct.
Yes, we can absolutely say that.
David Gerard (20:01)
I might get pushback
on that one, but I think that I'm going to reasonably assume it.
BJ (20:06)
You know, it's funny though, because I look over at the UK and I remember the thing with Liz Trust. And I remember, I remember everyone saying, OK, this person should not be here. Let's let's let's fix this. And the system seems a little more resilient as opposed to us where we can't there's no real mechanism. I mean, on paper, we have paper waves that you can unload the president. But in reality, it's designed in such a way that you can't. And that was, you know, that's going back to.
the way the country was formed. But I always feel like the UK and Canada and Australia have more of an opportunity to self-correct as opposed to us. Is that right? do you think, is that?
David Gerard (20:48)
Somewhat. I mean the same party stays in power. They just play musical chairs with who's in the hot seat and Rishi Sunak wasn't actually much more competent. He was just like basically it was the arse end of a clapped out government and it was going to change the next time around and then it changed to another government that after six months was already looking sort of clapped out and idealist. So it's great.
BJ (21:00)
That's right.
David Gerard (21:18)
Yeah, you can repurpose all your old Soviet Russia jokes for the UK these days, basically. So yeah, I don't recommend having a Great Depression II if you can avoid it. I think that not having it happen would be good.
BJ (21:28)
Yes, yes, agreed.
Yes, I mean, yeah, the ideal situation is always, you we have the New Kings, β which was the largest gathering in American history, just this year, and I'm sure it is gonna be even bigger ones. β
Sorry about that.
David Gerard (21:48)
Did that,
sorry, did that mute? I hit the mute button.
BJ (21:51)
Oh, it's OK. You cut out for just a second. That's no problem. I can work around that. Let me steer us little bit because I just am watching our time. I want to ask you about this shirt that you have in every episode where it says, you must be doing it wrong. Because this is something that the AI people say a lot. So I was hoping you might be able just to touch on that for people.
David Gerard (22:00)
Yeah.
So I used to do these little Bitcoin stickers. I still have them, β which say, it can't be that stupid. You must be explaining it wrong. When something is just very, very, very dumb and you explain it in precisely how dumb it is and people just go, that can't be right. And it is right. And it's very hard to explain a deeply stupid thing when it's very deeply stupid. And then a
BJ (22:23)
Okay.
David Gerard (22:44)
Guy on Macedon who uses the name burgers, mcslopshot came up with a variation. can't be that stupid. You must be prompting it wrong. So I immediately adopted this. He's going to thank you on the about page.
BJ (22:59)
I love it because I remember when this stuff was first coming out and I was still working in a SaaS business and that's what we kept getting told by the salespeople. I had no control over what was purchased and what wasn't. It was just one of those, hey, we're doing AI. But they said that a lot or some variation of you're doing it wrong. The system works great. And let me ask, in your...
In your observations, we know that there's a few use cases where this stuff, with a large language models can be useful, but on the whole, like, it's making us less productive. Do you see that?
David Gerard (23:37)
Uh, more or less they're sort of useful things that they aren't actually very useful for, are absolutely not the best thing for. Like the real use case for generative AI is toys. It's for messing around with. Like, I honestly feel the height of the GPT series was GPT2 because it was coherent enough to read it and it was broken enough to be interesting in its brokenness.
BJ (23:44)
Right.
Yes.
David Gerard (24:06)
You can have, it was fun. It was a fun computer to mess with, or even doing little memes with Dally Mini on Twitter or whatever. You know, it's silly. β But making it the cornerstone of our entire β economy and business is blitheringly stupid. It doesn't work. Even for the jobs it does do, it doesn't do them well. Translation, yeah, does translation because...
BJ (24:31)
Yes.
David Gerard (24:34)
It's based on, on transformer and transformers were invented for translation. β transcription, same thing. doing OCR. Yeah, it does that a bit, but again, more, a better model, more tuned. We'll do a much, much better job, much more efficiently. You know, like there's a little transform model inside Firefox that does the language translation and that you, that does not take.
an entire data center running chat GPT to serve it runs right there on your computer and it runs so fast you don't notice. You know, there's much more efficient ways to do all of this. The closest I've seen to a use case for these things is I speak to second language English speakers who use it to clean up their grammar. I, the real problem there is that they're dealing with
BJ (25:13)
Right.
David Gerard (25:29)
people who are looking for an excuse to be a bit racist and will use grammar irregularities as an excuse not to listen to you. That's a serious problem in academia, for example, you know, it's just a way to discriminate against people, but it's an excuse. So they try to get around the excuse and the hazard there is that your English doesn't get better and it changes your meaning and turns your writing into something that feels a bit like sludge.
And for someone who is a good reader and can tell the writing style from another writing style, it looks like slop and you don't want to look like slop. So it's fraught with hazards, even for that is being its sort of use case. The main thing it offers is that it's really, really, really convenient. Open the chat bot, put it on a picture, put in some words, put in a recording. It'll just do it. That's a good interface to a very
fat and wasteful thing that is sort of mediocre at the actual job. But you know, I mean, I can see why people use these things. The science fiction idea of a computer that talks is powerful. You can see why when chat GPT came out, people were absolutely blown away. This is amazing. We have invented at last a computer you can have a conversation with. That is something that's actually remarkable as a demo.
You know, it, because it turns out our talking computer is a lying idiot and that can't be fixed because it's inherent in the structure of how a large language model works. And this can't be fixed and it can't be fixed. And it's like, there's no better version of it that doesn't do this because it's inherent. Because facts are not a data type. That's the whole point of the transformer model is you don't care what the data type is.
You don't care about the facts. just go for correlations and that worked. It's unusually effective. But yeah, so I, I try to distinguish between machine learning, which works and generative AI, which sort of doesn't. The hazard with ML, and this is relevant to be on topic vaguely for the first time in 25 minutes. We, you have to think about machine learning because it works.
BJ (27:26)
That's right.
you
David Gerard (27:55)
It works well enough. You can use it to abuse people. You can get the robot to be racist for you, that sort of thing. So that's a different AI problem. It's an AI problem. It's a different one. The generative AI can do that, but also it doesn't work reliably and it can't work reliably. And we fully understand why it can't work reliably. And it's, it's nuts. So
But even then that distinction is sort of like between things that work and things that don't because quite a lot of good ML systems are transformers and LLMs are one step on from there. I would think in terms of what the thing is doing and also what it's for, that's the big political question because you can't say, but as a technology, does it work? You have to consider the political questions like.
BJ (28:33)
Bye.
Yes.
David Gerard (28:48)
Who is doing this and why are they doing it? Well, the answer is they want to fire everyone. You know, they've said this very loudly repeatedly for years. They want to make your life worse in material ways, but they'll give you a chat bot and they β tell you you can replace your brain with a clockwork mouse. And I strongly suggest you don't do that because you'll it'll make you dumb. If you want to be a dumb ass, this is a very good way to do it. So it's very fraught.
It helps that even if you do consider it just purely as a technology, it's sort of bad and stupid. You know, like we had people marketing Claude code, the Anthropic computer code writing bot so hard. And then the Claude code leaked and we look at it and was a pile of the most abject trash you've ever seen on a computer. And most of it was developer fan fiction where they were yelling at the computer to please not screw up this time.
And apparently Anthropic use this all the time. And this explains a lot about software reliability these days. And anybody who says this is an amazing new paradigm, it is the best thing ever. I don't know what to tell you.
BJ (30:02)
Do you, let me ask you before we wrap up, are you familiar with the story of Duke Nukem Forever? Okay, so.
David Gerard (30:08)
No, no familiar.
I know that it was an eternal project that never got finished. Or maybe it did, I forget.
BJ (30:12)
Yes. Well, it
did, but it was terrible. And it reminds me because, you know, for people that don't know, Nukem was really popular in the 90s. was like, was a doom knockoff essentially. β And it was very popular. And they kept saying, yeah, there's a sequel coming for nearly 20 years. And it just didn't happen. And then when it finally came out, it was awful. And I just can't help but feel that way about a lot of what we talk about with AI.
David Gerard (30:18)
Fantastic.
BJ (30:41)
It's just that we've been hearing about it and it's coming and it's gonna take your jobs. And now it's kinda here and it just does not at all seem like it's ready for prime time.
David Gerard (30:52)
It's, it really isn't, it does what it does. it, the large language models, there's been no paradigm changes in four years. You know, we have the same sort of thing we had in late 2022. It's a bit better. They, they make it better by doing a lot more computation where instead of running like 10 tokens through the machine, like a token is a word fragment. It's more or less the unit of doing
BJ (31:03)
Right, that's right.
David Gerard (31:21)
chatbot computation. Instead of running 10 tokens, they run a thousand tokens because of the bot checking and checking and back checking itself and so forth. And it's stupidly inefficient. And it's a technological S curve where everything starts off slow and then it accelerates so fast and eventually it levels out. We are well up into the top of the leveled out bit and the chatbots are sort of like
They're more or less the same and slightly better in some ways, maybe subjectively. I don't know. It's all vibes. And they, they have benchmarks, which they wrote themselves from marketing and it's just terrible.
nothing's really improved. You know, you'll have people who'll say, I think it's improved. It's done wonders for me. It's changed my life. Which means that what happened was they got the bot to do something right once and it one-shotted them. It just, it blew their minds and they'll just do anything to get back that feeling. And β it looks, very, it works very like gambling addiction, like every mobile game you've ever used.
BJ (32:11)
you
That's right. β
Yes.
David Gerard (32:33)
where they hook you with variable rewards. You get a reward. You don't get a reward this time. Better pull the lever again. Yes. And you you try to remember the wins and forget all the failures. And the main thing is keep spending money because the house always wins. Except in this case, the house is anthropic and open AI and they just set the money on fire and they are going broke, but they're not broke yet.
BJ (33:00)
Not yet, but we suspect they will be. We've been in that to this in future. Do you have time for one more question? Okay. I wanted to ask you real... Yeah, I'm good. β I really wanted to get in one question about Anthropic and Hexeth and dropping missiles because Hexeth had made a big show of wanting to take Anthropic through one method or another and essentially because he wanted their source code.
David Gerard (33:04)
fervently, fervently.
Yes. You have to edit all this together. It's up to you.
BJ (33:30)
And he kind of backed off a little bit, but now it's back in the news that Anthropic is claiming they've got, β forgive me, I can't think of what it's called, but they have a new model, which they're saying they won't release because it's too powerful. And so I'm just curious about this. Right, right, I'm sorry.
David Gerard (33:44)
They always say that. Sam Altman's been saying
that since GPT 2 was going to destroy the earth.
BJ (33:51)
It reminds me so much of social media, right? Like it reminds me so much of like the Gary Vaynerchuk's loudly saying, my God, this is going to change your life and you need to get in and cash in on your passion. But in terms of the military usage, just curious to hear your insight into that because there's some concern that that anthropic that may have been involved in dropping a bomb on the girls school in Iran. And I'm just curious to hear your take on.
sort of the military application of this stuff.
David Gerard (34:20)
β so firstly, they're using the chatbot. It's dumb as hell. β it can do some processing of data, but if you rely on the results in the event of whether, Hmm, who are we going to kill today? Then you're a blithering idiot. And so we had come to Pete Hexeth who clearly believes the chatbot is the best thing ever. He is a massive, massive Claude fan. thinks Claude is brilliant. It's a genius.
BJ (34:24)
Right, right.
David Gerard (34:49)
they're buying in Gemini and they're even buying in chapter GPT. They're thinking you're buying in GROK I think, but he just loves Claude the best. And that's why, and they don't understand how to do a good deal with people. So what they do is they try to browbeat them in and completely dominate them. And I'm not sure that's working on the other, but you know, if you, the anthropic connection with the bombing is that Palantir sold
defense a system that used Claude. So that's the anthropic connection there. it's, it's just so stupid. You will get wrong answers that are bad and you'll blow up a girl's school instead of the, β installation that was there like 10 years ago. And I forget the details. mean, I'm probably completely wrong and I'm sure someone will come back at me.
BJ (35:25)
Wow.
No, that's essentially, that was
essentially it. Like it was just old, old Intel was there was a missile site there, which later became a girl's school and they even fenced it off. And we saw from satellite imagery that there was like clearly like a fence or wall separating the two of them.
David Gerard (36:03)
Yeah. And it was just ridiculous. And if you use this stuff that you know gives wrong answers, that's on you. Like don't do the stupid thing. But Hakeseth is convinced that Claude is just brilliant and he wants to use it for everything all the time. He wants more AI. AI is the pay forward for the Department of War, the modern warrior. And
BJ (36:15)
Absolutely.
That's right.
David Gerard (36:33)
He's dumb as a rock, but again, I think that's an objective fact at this point.
BJ (36:34)
Yeah, that's, mean, that's a-
Yes, β that is something I can say for sure. He is actually as dumb as a rock. β David, this was great. Tell us, how can we support you? Because like I said, really think that you're doing, between you and Ed Zichron, β there's just not a lot of critical coverage of this stuff. So how can we support you? How can we get the word out about what you're doing?
David Gerard (37:02)
Well, the YouTube is called Pivot2AI and you'll know you've got it when you've got the little paperclip icon. The website is called pivot-2-ai.com and the website and the video channel and the podcast all link to each other. I'm davidgerrard.co.uk. That's my website and that's also my Blue Sky handle and
I'm posting there approximately 16 hours a day as one does. And β yeah, and I have a Patreon. If you like my stuff, me money.
BJ (37:42)
And I encourage people to do, you know, just speaking as an independent journalist, I know that there is no more media ecosystem in lot of respects. So if you do find someone whose work you enjoy, such as David's, please, please, please, you know, I know you have a Patreon. β So make sure you, if you can, I know it's tough out there, but if you can, please do throw some dollars David's way, because he is doing fantastic work. Let me press stop here. Just make sure we have fully uploaded.