@SubstantialFrivolity's banner p

SubstantialFrivolity

I'm not even supposed to be here today

5 followers   follows 0 users  
joined 2022 September 04 22:41:30 UTC
Verified Email

				

User ID: 225

SubstantialFrivolity

I'm not even supposed to be here today

5 followers   follows 0 users   joined 2022 September 04 22:41:30 UTC

					

No bio...


					

User ID: 225

Verified Email

It's been good for that sort of thing for a long time. In fact it's one of the only things that even I, who still doesn't think LLMs are very good at programming, use them for. When the code is short and trivial such that you can look at it and determine in seconds whether or not the model screwed something up, it can save you a lot of time. There are many such tasks which I would be just as fast at, but there are some (like writing bash scripts) which would take me several times as long due to having to look up syntax and such every step of the way. In those cases the LLM is a serious win.

I would say Braid's biggest plot sin is having a pretentious plot twist hidden behind getting the stars in each world, one of which requires hours of real time waiting for a cloud to cross the sky.

If fear of mass shootings is not considered "unreasonable neurosis", fear of false rape claims can't be either.

I wouldn't assume that @Skulldrinker doesn't consider fear of mass shootings to be unreasonable neurosis. I certainly do. The vast, vast majority of people will never be caught up in a mass shooting. It's silly to worry about them.

If he can't follow the rules, he should be banned. It's not rocket surgery. I don't care how interesting someone's posts are (and Dase didn't even make interesting posts IMO), the rules should apply to all.

Based on the use cases you mentioned, I guess I can't really say much because it seems like our situations are too different for me to really relate. I spend minutes on email every day, if that, and my bills are almost all paid with automatic payments so I can't miss them. So, it wouldn't make much sense for me to use any additional tool to optimize that, because the level of effort is so low already that it doesn't bear mentioning. It sounds your situation is rather more complex than mine, though.

Giving free access to my computer for wetware systems happens pretty regularly. My wife has all access to my digital life. The main impediment for her is just her time and willingness to read all the shit I write.

I agree that humans are not deterministic (I'm a believer in free will). But unlike an LLM, a human is capable of understanding things and reasoning about them. A human is also, unlike an LLM, capable of learning and responding to incentives. Your kids might hose your computer (they are kids, so it's to be expected they will make dumb mistakes), but they will learn. They can also be steered away from some things by laying out consequences. All of this makes humans lower risk to give free rein than an LLM.

I didn't provide many initial details. That was intentional. I wanted to see how people reacted to the basic thing without prompting in a particular direction.

It seems to me that asking for advice without giving details about the use case is just going to cripple the advice you get. That's why I asked for more info (though it turns out I can't give you decent advice for your situation anyway, c'est la vie). But to each his own, of course.

It would help if you articulated what, exactly, you aim to accomplish, as well as what your reservations are. I personally wouldn't, but I also don't find LLMs to be especially useful for the vast majority of things I do. I certainly wouldn't give them access to my computer, if that is what you're thinking of doing. Giving a stochastic piece of software with no understanding of what it is doing access to your computer is a pretty foolish decision, IMO.