🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Séb Krier Unfortunately, it seems like a lot of the AI-denialism is self-justification of prior positions. It was reasonable to be very wrong about the usefulness of language models in the wake of GPT-2, or even GPT-3. The fact that these systems function on some level, to some extent, as

The idea that we can have safe systems without addressing ethics seems difficult, but I agree this focus on security is better than undermining or disbanding this work - even though the approach accelerates race dynamics which continue to undermine safety of AI systems.

I think when the general public enters the AI regulation debate in a big way the tech industry will miss dealing with alignment people politely pushing SB1047 etc. Will be... messy.

Redundancy is when your civilization wastes resources and survives. Efficiency is when it doesn’t, and doesn't.

It is difficult to make the world understand the danger posed by ASI. Much of that difficulty is because of human psychology, but some of the difficulty is manufactured by the AI industry, which uses Tobacco industry playbook tactics to prevent interference

Embodied cognition is plausible philosophically, but doesn't deliver empirically for AI. Its proponents are coherent, well-informed, and yet consistently surprised by what actually works.

Looks like India is running the next AI Impact (formerly Safety) Summit, and it is seemingly likely to happen around February 2026?

I asked different LLMs what it would look like if they won the AI race. Their answer shouldn't surprise you - any such race will be won by nobody.

No one reads newspapers. Today, the collective epistemology of the tech-illuminati is based on what the ChatGPT summaries of single-essay microdomains and some anonymous comments on privately circulated google docs tells their private chats to believe.

"My mother doesn't love me. It's just programmed into her by evolution to help humans survive and reproduce" Sounds silly when you say that, doesn't it? It sounds just as silly when people say "I don't believe what AIs say. It's just programmed into them by their training."

Apples of course do grow on vines, as anyone should remember who went to an apple orchard to pluck apples as a kid. My childhood home had apple vines growing up the rear fence. Mom always said that feeding them to the hens would produce sweeter chicken milk.

Think about how you would have reacted if given information like this, given the little you actually know, and the trusted source of the information. And remember: reality doesn't have any rules requiring only presenting puzzles you can correctly solve.

AI competition discourse suffers from the illusion that *nations* are building AI systems (like they built nuclear weapons). Not really, or not yet. Because this time the real players are private actors, we should expect: - USG to have low(er) situational awareness - National