Mark Zuckerberg

So we built a lot of complex systems to moderate content. But the problem with complex systems is they make mistakes. Even if they accidentally censor just 1% of posts, that's millions of people. And we've reached a point where it's just too many mistakes and too much censorship. The recent elections also feel like a cultural tipping point towards once again prioritizing speech.

So we're gonna get back to our roots and focus on reducing mistakes, simplifying our policies, and restoring free expression on our platforms.

So we're gonna get back to our roots and focus on reducing mistakes, simplifying our policies, and restoring free expression on our platforms.

So we're gonna get back to our roots and focus on reducing mistakes, simplifying our policies, and restoring free expression on our platforms.

After Trump first got elected in 2016, the legacy media wrote nonstop about how misinformation was a threat to democracy. We tried in good faith to address those concerns without becoming the arbiters of truth. But the fact-checkers have just been too politically biased and have destroyed more trust than they've created, especially in the U.S.,

After Trump first got elected in 2016, the legacy media wrote nonstop about how misinformation was a threat to democracy. We tried in good faith to address those concerns without becoming the arbiters of truth. But the fact-checkers have just been too politically biased and have destroyed more trust than they've created, especially in the U.S.,

Is there a current set of methods that seem to be scaling very well? Right. So. With past AI architectures, you could kind of feed an AI system a certain amount of data and use a certain amount of compute, but eventually it hit a plateau. And one of the interesting things about these new transformer-based architectures over the last five to 10 years is that we haven't found the end yet.

Is there a current set of methods that seem to be scaling very well? Right. So. With past AI architectures, you could kind of feed an AI system a certain amount of data and use a certain amount of compute, but eventually it hit a plateau. And one of the interesting things about these new transformer-based architectures over the last five to 10 years is that we haven't found the end yet.

So that leads to this dynamic where Llama 3, we could train on 10,000 to 20,000 GPUs. Llama 4, we could train on more than 100,000 GPUs. Llama 5, we can plan to scale even further. And there's just an interesting question of how far that goes. It's totally possible.

So that leads to this dynamic where Llama 3, we could train on 10,000 to 20,000 GPUs. Llama 4, we could train on more than 100,000 GPUs. Llama 5, we can plan to scale even further. And there's just an interesting question of how far that goes. It's totally possible.

that at some point we just like hit a limit and just like previous systems, there's an asymptote and it doesn't keep on growing, but it's also possible that that limit is not going to happen anytime soon.

that at some point we just like hit a limit and just like previous systems, there's an asymptote and it doesn't keep on growing, but it's also possible that that limit is not going to happen anytime soon.

And that we're going to be able to keep on just building more clusters and generating more, you know, synthetic data to train the systems and that they're just going to keep on getting more and more useful for people for quite a while to come. And again,

And that we're going to be able to keep on just building more clusters and generating more, you know, synthetic data to train the systems and that they're just going to keep on getting more and more useful for people for quite a while to come. And again,

It's a really big and high stakes question, I think, for the company is because we're basically making these bets on how much infrastructure to build out for the future. And this is like hundreds of billions of dollars of infrastructure. So, like, I'm clearly betting that this is going to. keep scaling for a while.

It's a really big and high stakes question, I think, for the company is because we're basically making these bets on how much infrastructure to build out for the future. And this is like hundreds of billions of dollars of infrastructure. So, like, I'm clearly betting that this is going to. keep scaling for a while.

But it's one of the big questions, I think, in the field, because it is possible that it doesn't. You know, that obviously would lead to a very different world where it's I mean, I'm sure people still figure it out eventually. You just need to make some new fundamental improvements to the architecture in some way.

But it's one of the big questions, I think, in the field, because it is possible that it doesn't. You know, that obviously would lead to a very different world where it's I mean, I'm sure people still figure it out eventually. You just need to make some new fundamental improvements to the architecture in some way.

But that might be a somewhat longer trajectory for, OK, maybe, you know, the kind of fundamental AI advances slow down for a bit. We just take some time to build new products around this.

But that might be a somewhat longer trajectory for, OK, maybe, you know, the kind of fundamental AI advances slow down for a bit. We just take some time to build new products around this.