We're going to look back at the second Grump admin as what happens when society enthusiastically embraces ego-stroking hallucinations - from "magic computer" LLMs, hollow TV personalities, and of course good old combative dementia.
> We’ve mentioned Cavanaugh here before, for the time when he was head of the US Institute for Peace, and Elon and DOGE falsely labeled a guy who had worked for USIP a member of the Taliban, causing the actual Taliban to kidnap the guy’s family.
My guess is this will garner attention for use of AI — that's where my attention went as well initially. But there's another layer to this, which is whether a grant should be terminated just because it pertains to DEI, regardless of AI being involved or not.
My guess is you couldn't get a roomful of experts to agree on what "DEI" means; I doubt AI could do better, and even if it could, I'm not sure I'd want that to be the determining factor about whether it would get funded. To the extent it was, I'm not sure it would be a bad thing.
I wonder what the economic cost of DOGE basing policy entirely on whether something is DEI or not. Talk about cutting off your nose to spite your face.
>"Begin with ‘Yes.’ or ‘No.’ followed by a brief explanation. ”"
GPT models generate tokens from left to right, they are causal. That prompt causes the model to lock in to an answer and then generate the explanation after the fact. This is why you can sometimes see the failure mode "The answer is X because the answer can't be X so the answer is Y"
Asking for the Yes/No to be placed at the end would put the CoT before and generate 100% objectively better results.
I used to think prompt engineering was a bullshit term like you don't need to be trained at all to use this thing. But apparently you need to a little bit.
So if the idea alone that an application is fed into chatgpt isn't dumb enough, consider that they failed to even use chatgpt correctly, which apparently is a thing.
I wonder if the same thing happened with--or is happening at--NSF? I know researchers that did not get funding for quantitative ecology fellowships or grants. After back channeling with program managers, it seems that using "diversity"--as in the quantitative ecological measures, metrics, or derived functional values--may have flagged proposals to be rejected.
19 comments
[ 2.0 ms ] story [ 35.2 ms ] threadSorry for the OT, but... what on earth?
My guess is you couldn't get a roomful of experts to agree on what "DEI" means; I doubt AI could do better, and even if it could, I'm not sure I'd want that to be the determining factor about whether it would get funded. To the extent it was, I'm not sure it would be a bad thing.
Management: "We need to do X"
AI does X
Management: "It's not working"
AI: what do you mean? It does exactly what you asked.
Management: "I wanted it to do Y, and that's how you do it" (with Y having nothing to do with X whatsoever)
AI: ...
Management: I'm hiring the developers back ...
>"Begin with ‘Yes.’ or ‘No.’ followed by a brief explanation. ”"
GPT models generate tokens from left to right, they are causal. That prompt causes the model to lock in to an answer and then generate the explanation after the fact. This is why you can sometimes see the failure mode "The answer is X because the answer can't be X so the answer is Y"
Asking for the Yes/No to be placed at the end would put the CoT before and generate 100% objectively better results.
I used to think prompt engineering was a bullshit term like you don't need to be trained at all to use this thing. But apparently you need to a little bit.
So if the idea alone that an application is fed into chatgpt isn't dumb enough, consider that they failed to even use chatgpt correctly, which apparently is a thing.
https://en.wikipedia.org/wiki/Alpha_diversity https://en.wikipedia.org/wiki/Beta_diversity https://en.wikipedia.org/wiki/Gamma_diversity https://en.wikipedia.org/wiki/Zeta_diversity