Meta Muse Surges After Launch

Beth Lyons (0:00)

Hey, good Monday morning, everybody. So happy to see everyone. So happy to see Andy. Today is Monday, September 21st, and it is episode 816 816

Yes, if you've listened to me, it's the show, I am surprised. Every time I say a number because it just keeps getting bigger, which of course is what happens. Andy.

Andy Halliday (0:31)

I had the same experience every time I realized how old I am.

Beth Lyons (0:36)

Ooh, that's, yep. Okay.

So you are watching or listening to The Daily AI Show. I am Beth Lyons. With me in the studio today is Andy Halliday. Handy Andy Halliday. There we go.

Andy Halliday (0:49)

Good morning.

Beth Lyons (0:52)

We don't know if other people will join us. You will discover when we do as well. Andy.

So one of the things that I noticed was that the newsletters that came this morning, we're talking about the hack of OpenAI that we talked about on Friday.

Andy Halliday (1:09)

Yeah, we're always ahead of the game.

Beth Lyons (1:11)

We're like freaking, hey, we're not just regurgitating our reactions to stories that we read in the newsletters. Although we do actually, we're not just regurgitating, we're reacting with all of you because that's what this show is about, is like, hey, we're reacting in real time just like y'all are. But it's very cool when we cover the news before our favorite newsletters too.

Andy Halliday (1:38)

Okay.

So what caught my attention over the weekend was the reaction and the metrics around Muse, Meta's new personalized agent that's accessible and integrates with all of the different Meta platforms. So it overtook ChatGPT as the number one app on the US App Store, and that was just a week after its launch. And what I took away from the weekend was watching a gushing promotion, but unpaid promotion by Claire Vaux, who is How I AI and a really good influencer in the world of AI, who has tried all the different personal agents and really was very, very pleased and almost proud of what was possible with Muse as opposed to the others that she's worked with. So that's a big endorsement in my view. And then I also saw, I think, I forget in one of the newsletters this morning, there was a pretty complete comparison of each of the different agents, whether that's Claude Code or it's Codex or Muse, doing personal tasks. So it would have been done in co-work on Claude, for example, but doing personal tasks and a side-by-side comparison in which Muse was the only one that successfully completed the task, and each of the others failed in different ways. So I think the takeaway is that Meta, working diligently behind the scenes, is now I think coming out with something that is really impressive and is really, really, you know, tuned to consumer acceptance in a way that maybe the others aren't.

And I think that's really good news for Meta, because we've kind of cast shade on Meta for a long time, because of their failure to reach something akin to the frontier model's performance.

Beth Lyons (4:00)

So, I heard someone say that as well. There was a lot of conversation about, what's the other one that just came out? Inflection in, there's a...

Andy Halliday (4:17)

Gemini to keep track of.

Beth Lyons (4:20)

But, all right. I've tried to troubleshoot the thing.

Andy Halliday (4:25)

You're definitely streaming in landscape. Let me go on. Let me go on. So, that's Muse. That's news. News you can use about Muse. Do that. So, Meta's Muse. Try it. It's apparently very good. It's downloadable on the app store and it's also accessible through, I guess, the desktop. Now, this week, Xi Jinping, the premier of China, the grand leader of China, is coming to Washington and AI is a headline agenda item in that meeting, alongside trade issues, alongside Taiwan as an ongoing issue between the United States and China and Taiwan.

It seems that Sam Altman, Jensen Wong and Tim Cook are going to be part of a related executive dinner with Xi Jinping. All of this in the context of the discussions we had last week and the major news around the US promoting a no restrictions continuation of the development of frontier models, while the leaders of each of the main frontier models are saying, no, we have to slow down and we have to achieve some kind of agreement with China, particularly to get rid of the objection to the slowdown and containment of models and governance of those models and their development.

Dismiss the argument that, well, China is not going to slow down. So here's an opportunity for Trump and his advisors in the AI realm to achieve an agreement, but he's already kind of poisoned the possibility of that by saying, we're not going to stop. And that's his official position. Now he has, however...