Can AI Learn Good Judgment?

tech

Can AI Learn Good Judgment?

Plus: Dan’s attempt to clone Kate, a shortcut for turning demonstrations into skills, and the human goals machines still need us to set

TL;DR

AI can learn from diverse evidence like edits, screen recordings, or tool access.
The 'AI & I' podcast episode with Edwin Chen explores the future of human purpose in an AI-dominated world.
Edwin Chen believes AI will soon outperform humans in all tasks, shifting the human role to creation by choice.
The need for humans to set goals for AI is emphasized, as AI lacks intrinsic motivation.
AI models optimized for engagement may prioritize user retention over task completion efficiency.
The article also mentions ongoing experiments at Every with AI training methods.

Continua a leggere l'articolo originale

clem 🤗 reposted

MTS

SITUATION EXPLAINED: What even is a distillation attack?

@DanielleFong, founder of LightCell Energy:

"I understand you don't wanna get automatically hit by a bunch of subscriptions and then distilled by DeepSeek. I do get it."

"But you learned from the intellectual output of all of humanity. And we will be learning from the transcripts of what we work with with AI. This is inevitable. This is absolutely a bidirectional thing."

"Is putting all my transcripts and tweets into a vector database and having the AI check every turn for something relevant... is that a distillation attack? I would argue no, but it certainly is transferring information."

Yann LeCun reposted

Rohan Paul

Jensen Huang on "distillation"

On his new interview with axios, he was asked this question

"Should open source model companies be allowed to distill closed models"

"Distillation—learning from AI, learning from other people, and learning from other sources of knowledge, is fundamental to intelligence.

We are constantly learning from other people. I am learning from you through the questions you are asking, and you are learning from me. All day long, we are learning from one another. AI also has to learn from something.

The original AI models, whether they were open or closed, were trained on previously created knowledge from the internet. Now, AI is generating more content than humans. In a few more years, the internet could be 99% AI-generated content, and that content will have been created by some form of AI.

As a result, AI systems will constantly be distilling knowledge and intelligence from other AI systems. The fact that AI can learn is a good thing. We want AI systems to be intelligent because a smarter AI can also be a safer AI."

----

From "Axios" YouTube channel, (full video link in comment)

Yann LeCun reposted

Haiyu Wu

Learning causality from internet videos in latent space first, and then using RL to teach the foundation model how to act.

This approach is 30× cheaper than Gemini 3.1 Flash on pretraining and achieves a better result. JEPA is all you need!

However, actions are still learned during post-training. Figuring out how to automatically learn actions without action labels is still very important!

Yann LeCun reposted

Subbarao Kambhampati (కంభంపాటి సుబ్బారావు)

The Parable of the Neuro-Symbolic Drowning Man #SundayHarangue

In this well known parable, a deeply symbolic man, beset by the rising flood of activations caused by deep learning, goes up his roof and prays for help from his Neuro-Symbolic God.

When he eventually drowns in the activations and goes to the gates of AGI, he sharply upbraids his God for not saving him from the tide of deep learning.

God looks at him exasperatedly, and says,

*I sent you the LLM rescue boat that sailed over the activations with a symbolic (linguistic) I/O powered by the civilizational symbolic knowledge.

You looked away saying you are waiting for neuro-symbolic help.

*I then sent you a Verifiers/Harness helicopter rescue, that leveraged procedural symbolic civilizational knowledge in a "Generate-Test" fashion to further fortify the LLM boat.

You ignored that too, saying you will surely get neuro-symbolic help.

What the heck more were you waiting for exactly, my foolish child? 🤦‍♀️

Addy Osmani reposted

The Pragmatic Engineer

An agent can tell you it's correct, not that it's good. @addyosmani, author & director of engineering, on finding your edge over AI coding agents:

“I always go back to: what is alpha? My definition of alpha is advantage. What is the current thing that models are not very good at doing?

For software engineers, we often say that your alpha is in taste. Are we building the right thing? Is the thing that we are building actually good? Sometimes people will say an agent can tell you if it's good. I push back on that. An agent can tell you if a thing looks correct, if it's matching a spec. It doesn't mean it can tell you what's good.

Good can mean good from a user experience perspective, could be delightful, could be something that a person will actually want to come back to. And it is still sufficiently nuanced that it's going to take time for models to catch up to a point where they can replace that fully.

We tell people that judgment, verification, all of these other aspects continue to be important. But even if you say, maybe a year or two from now, models will catch up. We still need engineers to be answerable for these systems. That doesn't happen overnight. That happens when you understand a system, people trust you and you have that expertise.”