asey

This is complete malarkey - OpenAI is _not_ "straight-up channeling ideas from user input"

That's not quite right - Levent and Buckmaster did not actually have the Millennium Prize qualifying NS solution but something more limited. OAI invited Buckmaster to join and help rewrite the full solution paper but did not feel it was appropriate to invite an Anthropic employee to join as well - particularly given they were using internal unreleased models.

On the livestream (perhaps elsewhere?) you can watch the submissions and scores come in over time. The LLM steadily increased (and sometimes decreased) it's score over time though by the end did seem to hit a lacuna. You could even see it try out new strategies (with walls e.g.) which didn't appear until about half-way through the competition.

Could you give an example?

Those are ChatGPT models. The code-davinci-002 model is still available - they responded to community requests to keep it up.

Models like these don't see words as made of up letters but rather see the whole words (tokens) as a single entity. The result being they're not very good at creating novel (non-memorized) anagrams/palindromes and the like.

Are you sure that’s GPT-3? Doesn't match my experience of its outputs at all.

Of course, though in fairness parent was answering grandparent's specific question (and accurately in my experience)

But where is the apple? That is, where is the emergent phenomenon of the apple?

I think that would be a question of memory rather than having a conscious experience though I think it raises the question of whether consciousness can exist without some mechanism for memory even if only short term.

Hi Ben, could I ask the size of your team managing these clusters and development efforts?