The State of AI in Two Breakthroughs

After a very finance-oriented previous Leaders newsletter, today, we’re putting ourselves at the bleeding edge of this technology.

For that, we will cover research that answers two questions:

How does state-of-the-art frontier research look today? What are the Frontier Labs obsessed with right now?

How does the “efficient frontier” look today? What is the biggest recent breakthrough in doing more with less?

By the end of this read, you’ll have a better understanding of this technology than you ever thought you would have. If AI is about to transform our future, understanding it better than anyone else around us feels like a key competitive advantage.

Let’s dive in.

What the Frontier is Obsessed About

In frontier research today, the hottest topic right now is on-policy self-distillation, or OPSD, the four hottest words in San Francisco right now.

I know, it sounds complicated because it’s named to sound like it, but it’s actually not at all if you break it into first principles, which is what this newsletter is all about.

And to break down frontier research into first principles, we need to start humbly.

How do AIs learn?

Simplified much, all AI models follow the exact same process to learn: they make a prediction about what we want to learn, we measure that prediction against the ground truth (what it should have predicted), and we use the difference between the two predictions as the learning signal.

If the AI should have predicted 10 and it predicted 1,000, the difference is 900, a large error. If the next prediction is 500, the difference is lower, prompting the model to continue heading in that direction.

However, AIs never predict single numbers (or rarely); they almost always predict probability distributions. In other words, they don’t say “the answer is 10,” but “I believe the answer is 10 with 90% certainty, but it could be 12 with a certainty of 10%.”

This means the AI has responded with two numbers, not one, and associated probabilities, creating a distribution of possible responses.

We won’t get into this, but a purist of the noble art of statistics would tell you these aren’t actual probabilities because they aren’t calibrated. I won’t elaborate, I’m just hedging myself against the ‘statistics police’ to ensure they know I know. But I digress.

Notice the word “believe,” because I used it on purpose. By outputting a distribution, the model expresses its degree of uncertainty about its prediction; we are still forcing it to make a choice, but we allow it to indicate how “convinced” it is about the prediction.

Source

Distributions are great because they prevent “response collapse”: if you force the model to always choose one and only one option, creativity goes out the window because the model learns only the most likely option, not the others.

That’s the difference between a Large Language Model (LLM) that always responds “Let’s bake a cheesecake” to the question “What shall we bake today?” and one that responds with a more varied set…

Enjoyed the introduction? Subscribe or sign in to continue with this edition.