Passing the Turing test isn’t really the goal — e.g. it doesn’t try to fake a realistic amount of ignorance about the birthdates of US presidents. But if it were, it would pass here: many humans would tell you, confidently, that there are no cars in space.
RESEARCH
-
Challenge of incorporating all known trivia into training data
By
–
Expecting it to fully integrate every piece of trivia seems unfair — a lot of people would rate the first answer as reasonable. It’s fundamentally difficult to get training data that incorporates everything the model knows, rather than what the human demonstrator knows.
-
GPT-3 improvement shows not just pattern-matching, still hallucinates details
By
–
The point isn’t that it’s perfect, just that it’s not narrowly pattern-matching on this specific list of questions — there’s clear improvement vs. GPT-3 across many questions that contain false assumptions. It does still hallucinate details in other ways.
-
NeurIPS Rejects Sound Papers on Ethics Grounds
By
–
NeurIPS continues to reject scientifically sound papers on “ethics” grounds. How much longer will we put up with this?
-
NeurIPS hybrid conference format criticism virtual talks live posters
By
–
Apparently some people’s idea of a good hybrid conference is that all the talks are virtual and only the posters are live. Not kidding: that’s NeurIPS this year.
-
NeurIPS demotes sponsors to exhibitors over industrial concerns
By
–
NeurIPS sponsors this year were demoted to exhibitors, because we don’t like to soil ourselves with industrial ties.
-
Reinforcement Learning as Backpropagation Through Time
By
–
Reinforcement learning is backpropagation through time.
-
NeurIPS Auto Captioning System Mistranslates Conference Name
By
–
The NeurIPS auto captioning system translates “in NeurIPS” as “in Europe’s”, proving once again that the new name doesn’t cause any confusion.
-
NeurIPS Attendance Drops Significantly in 2022
By
–
NeurIPS attendance (real + virtual) is down by more than a third this year.
-
GPT-3 succeeds on trick questions, pre-training unchanged since 2021
By
–
No. It also succeeds on questions I’ve written myself that trick GPT-3, as in this screenshot. The pre-training data hasn’t been updated since 2021, and OpenAI specifically denies (in this thread) training on the Hofstadter/Bender questions.