wxnx

I'm curious -- in your view, what exactly is it strangling that could not be addressed by replacing google.com with a ChatGPT-like interface while integrating other Google free-by-ads tools nicely? Or is your point they wouldn't do that due to existing "siloed incentives"?

> A binary response vs rating is not related whether it learns confident or hedged tone.

Yes, I am aware. I was referring to the fact that at this point, user preference optimization is not done with RL, but with other techniques. I was unintentionally being pedantic.

> But there's an even more contrived reason the training set contributes. The vast majority of human writing is confident.

Yes, I think this has more to do with it than user preference optimization, honestly. "Valuable" text (i.e. text that produces a "good" model) for pretraining has the characteristic of being confident. Even models which are not optimized for chat (i.e. definitely no user preference data used to train them) exhibit this characteristic for medical questions (I know, because I've literally tested them for this purpose).

Preference optimization (or even RLVR) might play some small role as well, but it's kind of a "turtles all the way down" type problem.

Sorry, I understand now we're talking about different things. You're talking about preference optimization.

I was unintentionally being pedantic, because this isn't really done with RL anymore - it doesn't need to be. RL is now typically only used to train reasoning for tasks with a well-defined correct answer (that's what I meant by binary reward) - this is called RLVR (RL with verifiable rewards).

Preference optimization (training the model on user "this response is better than that response" type data) is more often done with something in the same family as DPO (direct preference optimization), which is decidedly not RL.

Your philosophical concerns are correct of course. And there's the added caveat that the models that most people are using are closed, so we don't actually know their training recipes for sure.

> They are always confident, because a confident tone ranks better in RL.

This makes it sound like RL rewards a confident tone -- in general, I don't think this is true (most RL is RLVR, which typically uses binary verification of correctness).

I say this because the real reason "they are always confident" is in some sense even more contrived. Training text where the speaker sounded more confident is more likely to contain a correct answer.

It is unfortunate that the only way to get through the regulatory process is copious $ - but it does work. If this project were better funded, it would likely have gotten through.

There is an entire political party representing something like half the population of the US dedicated to shrinking the regulatory apparatus, including the FDA. That doesn't sound like career security to me.

My perception is that the parent commenter is referring at least partly to "subcriminal" or nearly subcriminal behaviour. There is/was nowhere to "put" this kind of individual when they come from these settings, for a whole host of reasons, ranging from: something like "it's your neighbour's mentally ill kid, you really gonna call the cops for THAT?" to "what cops?".

School shooters and "crazy people acting violently" are probably incredibly rare by comparison. Sub/urban environments probably reduce the impact of the mentally ill on their surrounding neighbours in a whole host of ways, in fact -- in no small part because treatment is much less available in rural areas. (There's more of them in a smaller area and they're more visible and increasingly less criminalized, so people get the idea that poverty and health issues of that variety are modern, urban problems.)

Keywords in the literature around this include "community impact" or "community health", but it's not my area.

I really want to know this, too.

This is cool. I hooked up Llama to an open-source TTS model for a recent project and there was lots of fun engineering that went into it.

On a different note:

I think the most useful coding copilot tools for me reduce "manual overhead" without attempting to do any hard thinking/problem solving for me (such as generating arguments and types from docstrings or vice-versa, etc.). For more complicated tasks you really have to give copilot a pretty good "starting point".

I often talk to myself while coding. It would be extremely, extremely futuristic (and potentially useful) if a tool like this embedded my speech into a context vector and used it to as an additional copilot input so the model has a better "starting point".

I'm a late adopter of copilot and don't use it all the time but if anyone is aware of anything like this I'd be curious to hear about it.

Prior evidence suggests that sedentary behaviour increases blood biomarkers of cardiometabolic dysfunction.

They studied a hunter-gatherer tribe in Tanzania and found that despite having similar patterns of sedentary behaviour, their blood biomarkers of cardiometabolic dysfunction were much lower than those in industrialized nations. The authors posit that one reason for this is that sedentary behaviour in this group of individuals does not involve furniture -- rather, it usually involves a "deep squat", and the authors show that in this position the muscles are much more engaged than when someone sits in a chair.

This is consistent with evidence that breaking up periods of sitting with movement is good for you.

Their open-access paper talks about some evolutionary context for this hypothesis [1].

[1] 10.1073/pnas.1911868117

Eh, with one confirmed injury, I'm comfortable calling it "conceivable". The testimony of the injured worker in that case even reflects the fact that a spiked tree does not guarantee injury or death to the operator -- indeed, first there seems to be substantial damage to the machinery, and only then, a risk to the operator [1]:

> Alexander later filed a lawsuit against Louisiana-Pacific claiming that the band saw had been weakened from previous strikes with nails, but that he was forced to work with the saw or face dismissal.

Nope, you're thinking of regression coefficients, where you'd be correct that usually the null hypothesis is $\beta = 0$. In this case, what's being reported are odds ratios, so the null hypothesis would be that OR = 1.

The parent comment's point is that although the reported effect is significant at $\alpha = 0.05$ (the usual "95% CI" you mentioned), there are other problems that render their test of this hypothesis less than valid.

Any evidence for them being a fad? I definitely remember my parents' generation having dedicated one-on-one meetings with their managers on a regular basis, regardless of what field they were working in (skilled labourers, but none of them in tech).