Microsoft's AI Comeback, OpenAI Sites Side Quest, and Hollywood vs AI | EP103

Can Microsoft AI really catch up with OpenAI, Anthropic and Google DeepMind? Frank and Justin dig into Mustafa Suleyman’s Build announcements, the new MAI models, local AI, token costs, enterprise trust, and whether Microsoft needs the best model — or just one good enough to win inside big companies.

Plus: Anthropic calls for an AI pause while the race keeps accelerating, OpenAI Sites raises awkward questions about side quests and monetisation, and Martin Scorsese’s AI storyboarding work sparks a wider Hollywood debate featuring Reese Witherspoon, Nicolas Cage, James Cameron, Emily Blunt and more.

00:35 Is Anthropic serious about pausing AI?
04:29 Is Mustafa Suleyman reviving Microsoft AI?
06:30 What did Microsoft announce at Build?
09:37 Does Microsoft need the best AI models?
11:37 Are token costs making local AI attractive?
13:28 Can Microsoft spoil the AI IPO party?
18:33 Is OpenAI Sites just another side quest?
21:04 Is OpenAI Sites a monetisation play?
23:29 Is OpenAI Sites an enterprise headache?
25:53 Is Martin Scorsese replacing storyboarders?
30:17 Which Hollywood stars are backing AI?

► SUBSCRIBE
Don't forget to subscribe to our channel for more arguments

► LINKS TO CONTENT WE DISCUSSED

• When AI builds itself
• Microsoft and OpenAI broke up — now they’re ready to fight
• OpenAI Sites
• Martin Scorsese accused of ‘throwing artists under bus’ with AI storyboards
• The Big Hollywood Anti or Open AI Quiz

► CONNECT WITH US
For more in-depth discussions, connect Justin and Frank on LinkedIn.
Justin: https://www.linkedin.com/in/justincollery/
Frank: https://www.linkedin.com/in/frankprendergast/

Transcript

Justin: Hello. Good morning, good afternoon, good evening, and welcome to the AI Argument with myself, Justin Collery, hoping that things go faster and faster. Joined as ever by Frank Prendergast, who’s had a great week because it looks like sometimes some people want things to slow down.

Is Anthropic serious about pausing AI?

Justin: Late-breaking news, Frank. Anthropic have said that they want to stop the development of AI, or at least pause. Yeah, not gonna happen.

Frank: Fantastic news.

Justin: So let me bring you up to speed. So Anthropic released– They periodically publish these posts, blog posts, whatever, and in this one they said that, lookit, we are entering a period now where we are reinforced, where the models improve themselves, right? They’re self-improvement models, and that this is gonna greatly speed the development of new models, and that’ll be much, much harder to control, and therefore, they recommend that if everybody else stops developing the models, they will also stop.

And what they’ve said is that even if the development of AI models stopped for the next six or 12 months, there is such a huge capability overhang that the world would still change beyond all recognition, and therefore, it doesn’t in any way impinge on the vision of changing the world and making a better place for everybody.

But just continuing at the breakneck speed we’re going at is dangerous, and they’re prepared to stop as long as everybody else is prepared to stop too. What do you think of that?

Frank: I think that it is a very good idea. I think that it would be wonderful if it happened. I think that it’s good that Anthropic are saying it and putting it out into the ether and that it’s not just– I’ve forgotten the guy’s name, the guy from “The Social Dilemma” and “The AI Dilemma.” It’s not just people who might be seen as on the fringe saying, “You know what? We should slow down here.” It’s not just Bernie Sanders saying we should slow down. This is a very, very direct communication from one of the AI labs at the frontier saying we should slow down. But–

Justin: Smells of virtue signalling. Smells of virtue signalling to me. Totally.

Frank: I don’t know about virtue signalling. I think their heart is in the right place. I think they probably know it’s not gonna happen. We’ve had Dario Amodei– We’ve talked before about Dario Amodei being on stage with Demis Hassabis and the two of them saying, “Yeah, we’d love to stop. We’d love to slow down.” It’s not gonna happen. So–

Justin: Let me tell you exactly why I think it’s virtue signalling and why it’s not gonna happen. Also this week, we got word through Axios, and I think there was also an article in the Financial Times, where the NSA and the Pentagon were now using the Mythos model, that’s the one that Anthropic use, remember, and they’re using that to try and hack computers in other nation states, read China, right?

So the same company that are saying that, “Yeah, yeah, this is really dangerous,” their technology is being used by governments to attack other governments already, according to the Financial Times and Axios. And so even if we stop developing these technologies in America, the Chinese aren’t gonna stop because they’re already getting hacked by these models from the US.

Frank: Yeah. Look, I get it. I understand. I naively want us to do the right thing. The cynic in me accepts that you’re probably right, it’s not gonna happen. But at least this is putting it out there. I do think international cooperation is needed. I don’t think everyone’s just gonna stop on the kind of, “Hey, we read Anthropic’s blog post. Let’s all just stop now.” I think it’ll take huge international agreements and collaboration and cooperation, and we’re a ways off having that. We’re a long ways off having that. But fingers crossed.

Justin: Frank, they dropped two nuclear bombs and kept on developing those things for 20 years afterwards. It’ll take a big, big event for people to decide that this is dangerous and they want to stop developing it further.

Frank: Yeah. Yeah. And this is it. This is it. Well, what is that big event gonna be? Or will there be one? Who knows? Let’s–

Justin: Probably not. It’ll be fine. It’ll be fine.

Is Mustafa Suleyman reviving Microsoft AI?

Justin: Tell you somebody else who isn’t go- tell you somebody else who’s not gonna stop is Mustafa Suleyman. He’s had a pretty interesting week.

Frank: He’s back with a bang, huh? I mean, is it– he’s not–

Justin: No, he’s not.

Frank: Well, folks, this is why we call it the AI Argument. Well, look, I’m delighted to see him back. I’m really interested in what’s happening. So Microsoft had their Build Developer Con- everyone’s having a developer conference lately. They had their Build Developer Conference, and it seems like Microsoft are, I would say, back in the AI game.

Certainly, certainly communicating that they want to be back in the AI game. Their collaboration with OpenAI would’ve prevented them from doing this up until, I think, was it last year? Because they’re now building their own models, and they were not meant to chase AGI on their own while they were in that deal with OpenAI, which they renegotiated last year.

So now that they can do it, Mustafa Suleyman, who, when he joined Microsoft first, I think his main job was like, how do we make Microsoft Copilot not suck? And I believe he’s done a reasonably good job. I’ve actually been hearing better things about Copilot more recently. But now–

Justin: And that’s ’cause they’re using Anthropic and Coworker in the back end. They haven’t–

Frank: Is it?

Justin: Anything. Yeah, yeah. So when was the last time that Microsoft released a product that made you go, “Wow, that’s pretty cool”?

Frank: Oh, I can’t remember. Was it maybe never? I don’t know.

Justin: For me, DOS 6.22, which I’ve talked to you about before. It’s just 1994, I think it was, when I hopped on the bus to go in and buy that. So look at what– they call this MAI, Microsoft AI, I presume is what the M stands for. What have they released and what is impressive about it?

What did Microsoft announce at Build?

Frank: So yeah, they released seven models. They’ve got a new reasoning model, which is meant to be pretty good apparently. Now, interestingly, the benchmark that they used was that it’s about as good as, or maybe– yeah, it’s about as good as Opus 4.6 on the coding benchmark suite Bench Pro.

Now, that is pretty impressive, but Opus is up to what? 4.8 now. So it’s interesting that it is not at the frontier. And then they’ve got a lighter coding version, which is going to be cheaper and more efficient, et cetera. They’ve got an image model, which apparently is now number two on the image model leaderboards, and it’s even better than Nano Banana 2 for image editing.

So that’s pretty impressive. Then they’ve got a smaller version of the image model. They’ve got transcription models, so speech-to-text, and that is apparently beating Gemini and OpenAI’s models, and apparently it’s five times faster than any rival model. Again, very impressive.

They’ve got a speech generation model and a lower-cost, efficient version of that as well. And then they’ve got what they’re calling frontier tuning. So for the MAI Thinking 1, which is their reasoning model, you can basically fine-tune that using reinforcement learning on your organisation’s workflows and business logic and all the rest of it.

Justin: Yep. Sounds very impressive. Sounds very impressive, Frank.

Frank: And they’ve gone all in on agents like everyone else. So now they’ve got native support for OpenClaw. They’ve also–

Justin: That one is interesting.

Frank: It is, and also they’ve got their own model, what was it called? Scout. So that is built on OpenClaw and their Work IQ, which again allows you to train on your business, et cetera.

So they’ve basically got support for OpenClaw for the people who are willing to take risks out there, and then they’ve got the Scout, which is like the enterprise version, which is a little bit safer. They do have extra safety built into OpenClaw, et cetera. I think you’ll be interested in this as well.

They’ve got local models, so they’ve got something called Aion 1.0 Instruct and Aion 1.0 Plan. So the Instruct one is for the everyday uses, very small things like summarisation, rewriting, stuff like that. And then they’ve got Plan, which can reason, it can invoke tools and manage files, and they can be run locally.

They’ve also got a thing called Foundry Local, which is for more easily running open source models locally. And they announced hardware such as the Surface RTX Spark Dev Box, and apparently that can run a 120 billion-plus parameter model with a 1 million token context locally. So I think all of that together is pretty significant, I would think. What do you–

Justin: Okay.

Does Microsoft need the best AI models?

Justin: So, great. Okay, so a very impressive rundown, right, of all the different things that they’ve released. I think that they will not be the best models that you’re ever gonna use, and I’ll explain why in a second, and I don’t think it’s gonna matter. I still think they’re gonna do very well, unless they really screw up the rollout kind of like they did with Copilot and stuff like that, which they could, right? They’re Microsoft at the end of the day.

So why do I think that they won’t be the best models in the world, right? Mustafa Suleyman comes out, he says that he wants to be number one because he has to say it. Who wants to be number two, for God’s sake, right? I think the only company I’ve ever heard of, in fact, that ever advertised that they weren’t number one was a hire car company. I can’t remember which one. It was–

Frank: Oh yeah, that’s, yeah.

Justin: We try harder because we’re number two or whatever, right? That was sort of making a virtue out of the fact that they weren’t number one, but this is different. He has to say that, right? Because I think that’s marketing and the way it works.

But this is just one product in a whole load of other products that they have, and as they develop and they make these products, they have to figure out how they fit it into their overall ecosystem, and so there ends up being trade-offs.

Whereas OpenAI and Anthropic, and in particular Anthropic, have zero trade-offs to make. They just go, “We wanna make the best model, and in fact, for Anthropic, we’re gonna develop it to make coding really, really good.” And they focus on one thing, and therefore, that model tends to be better.

So that’s why I think their models will never be the best. That’s not gonna matter, though, because if I am just OCorp and I’m a big company, I probably use loads of Microsoft tooling already and I want to have, I want a very simple billing situation. It’s like Microsoft are the new IBM. Nobody really gets fired for buying Microsoft, right? It’s a known quantity. We deal with them already.

The security and compliance people will be much more happy to deal with a known quantity, all that sort of stuff. So I think so long as they make capable models, they don’t need to be number one. They just need to be good. They need to do what it says on the tin, and then I think they’ll do okay.

Are token costs making local AI attractive?

Justin: I think the other thing that’s interesting, I know you pointed out the smaller models which you say run locally. What I’m seeing is in large corporations, as the models become more capable, I talked about this before, right?

There’s a curve of capability, right? And two years ago, we were down here, so we were really, really, really waiting to get a better model so that we could do some more cool stuff. But now the curve has gone to the point where I think even as new models come along, they’re so smart, it’s really hard for the average person to notice the difference between, you know– Does the average person know the difference between a model that can act autonomously for 20 hours or 40 hours? No. Their tasks don’t extend to those time periods.

Companies are kind of the same, and they’re bounded by cost then at that point, right? So you have a whole load of models that can do the job at hand, and now cost becomes an issue. And if you’re renting a model, so you’re paying per token, what we’ve discovered is that’s very, very expensive.

And companies are– There’s been loads of examples. Who was it? There was one. Was it Airbnb, I think it was, blew through their token budget in four months. Everybody was just lashing out the tokens and whatever. Whereas if you can host a model, let’s say in an Azure instance, hey, Microsoft own that too, and then you can just run that instance and you can burn as many tokens as you want, that becomes a very attractive proposition.

Second thing is big companies often have concerns around data residency, and even just they don’t like data leaving their ecosystems if at all they can avoid it. So if you can just spool up a machine and run a model that’s good enough in your environment, well, that’s again, the security people will be delighted with that, and the accountants will too, because you’re not burning tokens.

So it doesn’t matter if their model isn’t the shiniest, best model in the world. It just has to be good enough, and I think they will be good enough.

Can Microsoft spoil the AI IPO party?

Justin: So Mustafa Suleyman may actually– I mean, is this– We have a number of– I know you’re big in the financial world. We have a number of big IPOs coming up.

Anthropic published, put in their papers for an IPO for how many? About a trillion dollars. And what’s X, Elon Musk’s company, has done the same. He’s hoping to go public. And OpenAI we know are hoping to go public before the end of the year as well. How does this play, I wonder, against that if Microsoft…

By the way, another interest- ’cause again, I know you love the money stories, right? An interesting thing happened this week as well, which was Google, how much money do Google have, do you think?

Frank: Oh, I have no idea.

Justin: More than you and me. More–

Frank: Yes, definitely, yeah. A little bit.

Justin: They’ve got a lot of money, right? And they make, I think, 30 or 40 billion every quarter, right? So they’re not stuck for cash, but they call it writing paper. They basically created debt for $80 billion this week.

And part of me is going, is all of this, the $80 billion there and these guys publishing the models which are just good enough, is this all– it’s big money stuff, and it’s to take some of the air out of the IPOs of Anthropic and OpenAI and X.

And it also plays back to– ’cause, I mean, I would share this sort of concern, right? You have these companies that are built on projecting revenues, which are based on bigger and bigger income from tokens as time goes forward. And for the reasons I’ve just outlined, companies are going, “Hey, hold on a second. We’re burning through these things really quick. We need to run these local models,” which hurts OpenAI and Anthropic and X.

And then also you have these Chinese models, which you can also bring in, and you can run them for about a 10th of the cost. And so it does put, I think, a question in people’s minds, is this incredible revenue growth that you’ve seen from Anthropic, is that really gonna continue on that hockey stick, or is it gonna start to do– that top of the S curve is gonna appear much more quickly?

So I don’t know. You’re a big Microsoft fan. Are you gonna use the Microsoft models?

Frank: I’m not. I’m on an Apple Mac, so I am not going to go anywhere near the Microsoft models, I would imagine. I do think you– I agree with a lot of what you said, and I think it’s very similar to what we talked about on the show quite recently in terms of Google, and whether it was the same for Google, that they don’t necessarily need to be at the absolute cutting edge. They just need to be good enough because they have the audience, they have the customers already.

But I do think that you might be underestimating Mustafa Suleyman somewhat, and I do think that he wants to be at the frontier. I mean, I’ve been really surprised at how quiet he has been in Microsoft since he joined there.

I know it’s not like he disappeared off the face of the earth or anything like that, but at the same time, he was a very prominent figure in AI. It was very much kind of like you would talk about Sam Altman, you’d talk about Dario Amodei, you would talk about Demis Hassabis, and you would talk about Mustafa Suleyman, and then he kind of disappeared from the public imagination somewhat.

And I think this might be him coming back, and I think this might be him freed from the shackles of OpenAI and going, “Right, come on lads, we’ve got work to do. We can do this.”

Justin: Maybe. Maybe. I don’t know. I don’t know. We’ll see. The other side of what I’m saying is that mind share is important too, so it’s tough.

I know I was talking to a friend of mine and they were at some sort of a conference over in the UK during the week, and everybody uses Claude Code. It was a developer conference, just everybody. Everybody uses Claude Code. So it’s Claude Code or OpenAI, and it’s very hard, I think, for anybody else to come into a professional organisation and sort of chip away at that.

Cost is the only big one I can see, but it would be difficult for Microsoft to get mind share in order to steal some of that away. It could be echoes of the Windows Phone about it. If anybody, does anybody remember the Windows Phone? Do you remember the Windows Phone?

Frank: I remember the name. I couldn’t even tell you what the Windows Phone looked–

Justin: Oh, actually, do you know what it is? Just for the, whatever, right? It’s just another example of a terrible product. So the iPhone–

Frank: I thought you were gonna take one out of a drawer there for a second.

Justin: I had one. I did have one. The Windows Phone was basically a phone that was supposed to compete with the iPhone. It was horrible. It was just horrible, and it had a big touchscreen on it.

It was really slow and had these tiles on it that you had to… And it just, it was another example of Microsoft just, you know, they’re just… They make an operating system and it’s, whatever, it’s good. They do a good job with the operating system, but all the rest of it.

Is OpenAI Sites just another side quest?

Justin: Anyway, OpenAI had a good week this week.

Frank: Why do you say that?

Justin: They released a thing called Sites, which I think is great.

Frank: Why? Well, first of all, tell us what Sites is.

Justin: Let me sell you the dream, right? And then you can tell me, ’cause this is actually more in your ballpark than mine.

So Sites is this new product which has been released by OpenAI, and what Sites allows you to do is to build a full-stack website and then host it with OpenAI. So you can go into your OpenAI, I presume it’s in your developer console. You can say, “Hey, I need to do a website for the AI Argument, and I’d like to sell some merch, please. And yeah, we’re gonna sell some cups and some T-shirts, and just can you magic that up for me, please?”

And it will create the website, do the full-stack thing, and then you just go publish, and your site is live, and it’s hosted on OpenAI. What’s not to like?

Frank: Yeah. It’s cool, but I don’t get it. First of all, right, here’s my big question. Are we back on side quests now?

So it wasn’t that long ago that OpenAI were like, “Oh my God, code red, code red. Anthropic are getting ahead of us.” And then shortly after that they were like, “Oh, we’ve gotta stop doing side quests. Google are getting ahead of us.”

Justin: All right, here’s–

Frank: Said, “Right, we’re gonna focus, we’re gonna focus on the core thing, the core product, the…” And now we have Sites? I mean, am I missing how this is part of the core product? Because personally, unfortunately, I wasn’t able to test it because it’s only on enterprise and business models, which I find really interesting because this strikes me as something that is aimed at the hobbyist vibe coder like myself, not the enterprise.

Because, I mean, you’re a developer. Are you gonna use this over the workflow that you’ve used for years, probably with GitHub and whatever hosting?

Justin: No, no. No, I’m not going to, right? But maybe other people will. But I think, in terms of the focus thing, right, I’m sure they have a Venn diagram somewhere in the office of things, and in one of the circles it’s got code, and in the other one of the circles it’s got money. And this does kind of sit inside that sort of Venn diagram of, we need to do code-y things and need to make money.

Here’s why I think it’s smart, and maybe it’s just a terrible implementation of what I think is smart, right?

Is OpenAI Sites a monetisation play?

Justin: So one of the big issues that the people in the wider world see at the moment is that AI is gonna eat Google’s lunch. So how is AI gonna eat Google’s lunch? Google makes loads of money from pay per click.

So you go onto Google, you type in what you want, the search results come up, and the top ones are paid for. But you only pay for them when you actually click on them. So somebody finds the AI Argument website, they’re looking for AI Argument merch, they click on the ad that we’ve put up there, and as soon as they click on it, they’ll take a penny out of our account or a cent out of our account, and they’ll go to our website.

And the problem is, now you don’t get the search results anymore. When you go into Google and you say, “Here, I’m looking for some AI merch,” it just gives you an AI answer, right? And it doesn’t have those paid-for clicks.

Except, right, what happens if instead you’re now on ChatGPT and you’re saying, “Hey, I really like that show, The AI Argument. I’d like to buy some AI merch, please.” And it goes– because it knows now, it’s hosted on their own infrastructure, and it goes, “Oh, that’s cool. Yes, you can buy it at this site here,” and you click on the link.

Now, that’s not an ad, right? So they can claim that they haven’t besmirched any of their morals by stuffing stuff into your feed, right? Because I paid for it. But as soon as you click on that link, they may well charge me a penny for serving the traffic of somebody visiting my website, and therefore, the economic outcome is the same.

It’s pay per click because they’re the ones hosting the websites, and that’s why I think it’s smart. Now, I think that Google are in a better position to capitalise this than OpenAI, but I think it’s a monetisation play. It’s a way to make money.

Frank: Wow, I don’t know. I don’t see– I can’t see it. I just can’t see it. And by mentioning Google there as well, you just raised something else kind of interesting to me, which is where are these actually hosted?

So I did a little digging around, and it seems to me that Cloudflare is involved somehow. So I didn’t really understand what D1 and R1 were, but these are basically how the databases for– if you’re deploying something that has a database or you need to upload images to or whatever, that’s how these are handled. And apparently these are Cloudflare technologies.

Is OpenAI Sites an enterprise headache?

Frank: So where are these actually being hosted? ‘Cause that, I couldn’t figure that out, but the Cloudflare terminology made me wonder, are they partnering with Cloudflare for this?

Which raises another can of worms for me, which is, earlier you mentioned about enterprises being nervous about where their data is sent, and that’s why there’d be interest in local models. Well, this is going out to the enterprise plans.

So now are you going to have a situation where somebody can hit deploy and company information is sent out to some hosting somewhere for some vibe-coded tool that somebody in the office has decided to do? This sounds like a big can of worms to me in terms of security and data privacy and–

Justin: Yeah, yeah. And you know my views on security and data privacy, but actually this instance I totally agree with you. This doesn’t look like an enterprise product at all. I think you’re dead right. Don’t know why it’s on the enterprise platform.

Frank: It’s weird.

Justin: This is way more like a mum-and-pop shop type thing where they just wanna have a website to gather some traffic and whatever. This is an easy way to set up a–

Frank: Yeah. Internal tools I would get, like maybe, you know, spin up a dashboard to do your presentation or whatever, but only internally. That I would understand for enterprise. No need to go out to GitHub, no need to go out and host it somewhere. I can see there’d be use cases, but only if it was kind of in a walled garden.

Justin: Yeah. Do you know what it is? Maybe it’s another one of these things, right? I remember getting very concerned about OpenAI when– Remember they released that agents dashboard thing? It was maybe six or eight months ago, right? It was a way to visually design agents and do stuff.

And I remember playing with it and going, “Yeah, it’s a bit meh.” Like, there’s work to be done here, and it’s very good, but it’s a bit meh. And then it came out that it was vibe coded in, like, three weeks before the presentation of it. And it’s like, really? And maybe this is another one of those things where it’s like somebody had a good idea, and they just keep throwing stuff to see what sticks.

Frank: Side quests. Side quests. They’re getting distracted by side quests, I’m telling you.

Justin: Focus. Focus, please. Okay, look at something that’s close to your heart.

Is Martin Scorsese replacing storyboarders?

Justin: Martin Scorsese is getting into a bit of hot water this week. One of my favourite directors, I love Martin Scorsese and his movies. He’s not the one that’s not allowed to go to Switzerland or leave Switzerland, is he?

Frank: Not as far as I know. No, I don’t–

Justin: Right. That must be a diff– That must be a different–

Frank: I don’t think there’s any Martin Scorsese controversy, thankfully.

Justin: Is it Polanski I’m thinking of, is it?

Frank: I would assume so. Yeah, I would assume so. Different kettle of fish. Scorsese, though, you know, Scorsese’s a funny one, just to get away from AI for a second. “Taxi Driver” is my favourite film probably of all time, Scorsese.

But bar “Taxi Driver,” I always thought I loved Scorsese, and recently I sat down and went through the films, and I was like, “Actually, I think ‘Taxi Driver’s’ the only one I–

Justin: What are you good at? I think that I only like the music in it. I really like that sort of the doo-doo music in it. It’s fantastic. What other movies has he made? Actually, just try his…

Frank: Goodfellas would be a huge one, and I always–

Justin: A good–

Frank: Well, I always think I like it, but then when I watch it, I don’t know that I actually enjoy it. I can appreciate it, but I don’t enjoy it.

Justin: Is that not a true story as well?

Frank: Oh, is it based on a true story? I can’t remember. I actually can’t.

Justin: I know. I say all movies are based on a true story.

Frank: But anyway, so he has– Martin Scorsese is now in some kind of partnership with Flux.

The relationship is actually a little bit unclear. I don’t know, are they paying him or what’s going on? But he was in a promo for them, and he’s exploring storyboarding primarily, using AI to storyboard quickly to get what’s in his head out to his team quickly and efficiently and in a better way than he could draw.

Which I think is an interesting point, ’cause a lot of people are saying, “Oh, he’s taking away storyboarders’ jobs,” et cetera, et cetera. I was reading just a minute ago that he storyboarded Taxi Driver himself. So yeah, I think it’s a perfectly valid use of AI.

I think it raises, it opens a can of worms around jobs and around creativity. But, I mean, if you’ve been doing your storyboards by hand and failing to communicate properly to your team, or as best you possibly could, now using AI to be able to go, “Look, this is what I’m talking about. This is how it would look photographically.” I don’t know, I think that’s a brilliant use of AI.

Justin: And just explain to me your head. So how does it… So if I wanna sit down or you wanna sit down and make a movie, right? You describe the scenes to a set of artists, and then they sort of draw them, and you sort of work with them to sort of say, “Yeah, it’s a bit this or it’s a bit that,” until it gets to the point where you’re happy with it. Is that how it works?

Frank: Yeah. Some storyboards are almost like a frame-by-frame, shot-by-shot comic book of the entire film. I mean, it depends director to director, but some directors will literally have a kind of a visual, a drawn visual almost like comic of the entire film before they start shooting.

Justin: And do you think that part of the concern then is that if we can do the storyboarding, it’s pretty obvious that you could probably just create the movie as well? Like, I mean, if I can storyboard it, doesn’t seem like a big leap.

Frank: I mean, yeah, I think there are all manner of fears. I think the biggest fear right now in terms of the Martin Scorsese story is, well, storyboarders are looking at this going, “Hang on, this is my job.” And I understand that and I feel for them, but I don’t know what we do about that.

I mean, unfortunately, this is the way technology and capitalism has always worked and–

Justin: Jesus Christ, what’s your hang-up with capitalism? It’s just technology, right? It’s qualitatively better than the previous way of doing it, right? If you’re a director, you don’t have to describe the thing. You can literally just go like this on your computer, and if you don’t like it, you can go and change it again.

It’s nothing to do with capitalism. It’s just progress.

Frank: No, but I mean, in terms of somebody’s job, it is very much to do with capitalism because nobody’s going to–

Justin: People in North Korea work too, Frank. That’s not capitalism. They also have jobs. They just get told what their jobs are. Anyway, let’s not go there now.

Which Hollywood stars are backing AI?

Frank: So I thought we’d do a quick check-in on the rest of Hollywood, right?

Justin: Oh, cool, cool.

Frank: And so myself and ChatGPT did a little bit of work earlier, and we came up with this. So–

Justin: Look at that. Okay, I love it.

Frank: So this is the big Hollywood anti or Open AI quiz, right?

Justin: Okay.

Frank: So Justine Bateman. Now, Justine Bateman is in here because she has been very vocal about AI in Hollywood, and she has been working with SAG-AFTRA to really, really…

Actually, what am I talking about? I’m gonna give this away if I tell you any more.

Justin: You are. Yeah, yeah, yeah. I mean, I was a big fan of Family Ties. Justine Bateman is one of my favourite actresses. I’d say she is on the open side of the argument.

Frank: Open. Oh, you are wrong. No, she wants absolutely no generative AI in the entertainment industry, period. She is probably one of the most anti-AI people that you could come across. Okay, who we’ve got next? We’ve got Reese Witherspoon. What do you–

Justin: Ah, Reese is great. Yeah. I’d say she’s all up for the AI as well, is she?

Frank: Open to AI? Final answer? Let’s go for it. Yes.

Justin: Woo-hoo.

Frank: Open to AI. So she says it’s a tool that we layer our own creativity onto and our humanity and our ethics. She says, “Let’s not be scared of it. Let’s dive in. Let’s lean in.” Good on you, Reese Witherspoon. Yes, I–

Justin: Very good. I like.

Frank: All right, Demi Moore, what do you reckon?

Justin: Demi Moore, now she’s getting on a bit. I’d say that she could be, she could be a bit of a Luddite now, the old Demi Moore. I’d say she’s anti, is she?

Frank: Anti, let’s have a look. No, she is open to AI. She said it’s here. There’s no point in fighting it. Basically, we’re gonna lose that battle if we try and fight the technology, which is again–

Justin: She’s smart.

Frank: Yeah, so we have to find ways with which we can work with it. That’s a more valuable path to take.

Great.

Justin: We should get her on the show to talk.

Frank: We should. Okay, Nicolas Cage. Nicolas Ca- oh, I love Nicolas Cage.

Justin: Nicolas Cage is one of the best and worst actors that have ever lived, right? And was Nicolas Cage… He’s been in some movies where he could do with some AI. I would say Nicolas Cage is pro-AI.

Frank: Pro-AI. Let’s have a look. No.

Justin: No.

Frank: Anti-AI. He says he does not want to let the robots dream for us because robots cannot reflect the human condition. Yeah, I mean, look, it’s a fair comment, but like we said, I just don’t think there’s any point in fighting it.

Justin: But not going to age very well. Guillermo del Toro, interestingly, interested to know that, but he was, is somebody that could do amazing things with AI. He has such weird and wonderful movies that are so, you know… Right, I could imagine that if he embraced AI, he would do– I’m gonna hope that he’s open, but I’m guessing that he’s anti ’cause he’s… I’m gonna say anti then. I’d say he’s–

Frank: Anti? That is correct. In fact, he said the other day, “Somebody wrote me an email and said, ‘What is your stance on AI?’ And my answer was very short. I said, ‘I’d rather die.'”

Justin: He looks a lot like Philip Seymour Hoffman in that picture.

Frank: Yeah, he does a little bit there. Yeah, yeah. Rather die. That’s about as anti as you could get really, isn’t it?

Justin: It’s a shame because he could do great stuff. James Cameron, I’m quite sure, is pro it actually.

Frank: Correct. Yes, he said the–

Justin: I’ve heard him talk about it.

Frank: The convergence of CGI and AI will unlock new ways for artists to tell stories in ways we could never have imagined, which again, yeah, I think that’s just fair enough.

Justin: Yeah, very good. Seth Rogen. These are all very mainstream people, Frank. Do we have any more alternative type directors on the list? Anyway, that’s not important.

Frank: I went as recognisable as possible. I think Justine Bateman is probably the least known name on the list, but I included her because of her importance in the argument.

Justin: Well, the reason I ask is, you know all internet technology developed because of porn, everything. The reason that you can stream TV is because of porn.

Frank: I have heard that.

Justin: Porn was at the forefront of every technological advance that we’ve had up until this point. And so, in fact, AI might be the first…

Anyway, that’s not important now. Seth–

Frank: Yeah. Let’s not open that can of worms right now. There’s probably a whole episode on that at some point.

Justin: Yeah, I’m up for that one. Seth Rogen, I would say is, I’d say he’s anti it actually.

Frank: Anti, absolutely. He is anti. He said, “Every time I see a video on Instagram that’s like, ‘Hollywood is cooked,’ what follows is the most stupid dog beep I’ve ever seen in my life. And if your instinct is to use AI and not go through that process, you shouldn’t be a writer because you’re not writing.”

Justin: Yeah, but that makes it the same as 80% of the other stuff that comes out of Hollywood, some would–

Frank: Yeah.

Justin: Maybe 95% of it. Ben Affleck, I would say that Ben, he’s actually one of my favourite actors and directors. I’d say he’s open to it. Yeah, I love Ben Affleck. I think he’s great.

Frank: Open to it. Yes, correct. In fact, he has his own AI filmmaking company called Interpositive, which was acquired by Netflix. And yeah…

Justin: Sorry. Who’s the guy that he, who’s the guy that he, who’s the guy that he does the movies with? What’s the other guy? Ben Affleck and Matt Damon.

There was a very interesting interview with Matt Damon. I don’t know if you saw it. He was explaining the money situation behind why don’t they make big, brilliant movies anymore, and he just breaks down the fact that people don’t go to cinemas anymore, that the cinema run is really short, and therefore the budget to actually make the movies is really small, and therefore you just end up with fairly bland, nondescript movies and whatever.

So that’s why I’m thinking they’re kind of business focused, the two of them.

Frank: I must look that up. Okay, Emily Blunt, last person.

Justin: Emily Blunt, ah, she’s got a… I mean, maybe she even has a strong opinion on it. I’m hoping that it’s open.

Frank: No, she’s anti it. Do you remember we talked about Tilly Norwood, that AI, one of the first kind of AI actors? Well, when that came out, she said, “Good Lord, we’re screwed. That is really, really scary. Come on, agencies, don’t do that. Please stop. Please stop taking away our human connection.”

And I think it was Emily Blunt, now I could be wrong, but I think it was Emily Blunt that I read just the other day, she’s got some kind of sci-fi movie, and she had to make– She had to speak an alien language, and she said that she was so terrified that AI would be used that she made all the noises vocally, and they just manipulated them in traditional audio equipment rather than using AI to make the alien noises.

Okay, last one, Natasha Lyonne.

Justin: Oh yeah. She’s great actually.

Frank: She is, yeah.

Justin: Great. Russian Doll was brilliant. I really–

Frank: First season of “Poker Face” as well, really liked the first season.

Justin: Yeah, yeah. Okay. I’m guessing, she strikes me as somebody who’s quite traditional. I’m gonna say she’s anti it.

Frank: Anti? No, she also has an AI film studio.

Justin: No.

Frank: And she said, “Yeah, we should use AI.” And she basically said there’s no avoiding it really, and so we should use it. So let’s see. You scored five out of 10. So–

Justin: It’s almost like it was a random chance each time, eh?

Frank: So there you go. Yeah, Hollywood, Hollywood, pretty mixed views on AI. Some big names getting involved in the conversation, and I think the ones who are just saying, “Look, there’s no fighting it,” are probably correct.

I did see Cate Blanchett has created a new company seeking to protect people’s likenesses, et cetera, from AI, which I thought was interesting. So she seems to be– she has huge reservations about AI, and she’s doing something about it, and I think she recognises it as a tool, but also sees places that need to be safeguarded and is taking action on that, so I thought that was interesting.

Justin: It is interesting. Somebody needs to do the other way around as well, though. I mean, if somebody who wants to promote their likeness in an AI. I would like– Do you know what? Did you ever watch the TV programme, since we’re doing this, what was it? The one where Rimmer was in it. He was a hologram. Oh, they were in a–

Frank: Oh, yeah, Red Dwarf, yeah.

Justin: “Red Dwarf.” I want to be the guy on the TV.

Frank: Who was the guy on the TV?

Justin: There was, you had Rimmer and you had, and then they had, like, a guy on the TV who would just pop up every now and then. He was like their AI in the–

Frank: Oh, yeah, yeah, yeah. Yeah.

Justin: So that’s it. So when they create a face, I mean, I’d like your face to be on there. If I’ve got a sort of face in the kitchen that’s gonna tell me how to cook mushrooms or something like that, I think it should be your face. So there should be a way to promote your face.

Frank: Well, Justin, if my face ends up on your fridge, I will be getting Cate Blanchett to go after you, so–

Justin: Very good. All right, Frank, have a great week.

Frank: Chat to you next week, Dustin. Absolutely.

Justin: Talk to you.