undefined | Better HN

0 pointsmyworkinisgood1y ago0 comments

Scientific computing is going to be overhauled with this. We have essentially standardized a way to approximately solve optimization problems. And people are now going to design methods to fit this kind of solution, just like people did with linear solvers.

Not to mention all the proof-writing that will become simpler with this optimization/searcher now.

0 comments

7 comments · 2 top-level

somesortofthing1y ago· 5 in thread

Sure, AI represents a substantial improvement over other heuristic methods in some areas. But that's a long way down from the "AI is going to be a permanent fixture in most people's day to day lives" claim that the tech industry is betting the farm on right now.

reaperman1y ago

I think you’re missing that “every” K-12 student is using ChatGPT for all their work right now. Yes, the state of education is in peril (with ChatGPT actually being pretty far down the pareto chart), but the generation coming up after us is absolutely growing up using LLMs the way we used calculators, and using it for everything.

We may not see universal adoption in people who are currently >30 but I think we will in the generations that are <25 now.

somesortofthing1y ago

But are they using it for anything else? Even GPT-2 was "good enough" for high school writing assignments. The fact that this power user class of young people hasn't found any other use cases for LLMs despite vast improvements in LLM capability doesn't reflect well on the merits of LLMs as consumer software.

reaperman1y ago

I think you may be having a failure of imagination. That generation will be creating additional tooling to use this wherever possible, and taking advantage of any interoperability they or their peers can hack onto any interface. Our generation will be hesitant but they will not be, and many of the upcoming generation will have a deep understanding of where LLM’s have appropriate vs. awkward application.

TeMPOraL1y ago

> The fact that this power user class of young people hasn't found any other use cases for LLMs

Why are you assuming they haven't? High school writing assignments and homework are the majority of the problems teenagers face daily, but they're also having fun with it, and why wouldn't they try it on new problems as they come along?

threeseed1y ago

Of course students are using ChatGPT. It helps them to write assignments.

But it isn't translating into better across the board test results and at least in Australia we would be able to tell because we have yearly standardised testing.

And so schools are looking at it as more of a form of cheating and simply moving back to in-person, hand-written tests.

pas1y ago

with what? how? there's already a lot of bad python/R code out there, how more of it will "overhaul" scientific computing?

> We have essentially standardized a way to approximately solve optimization problems.

.. what does this mean? we had simplex solvers before. do you mean things like protein folding prediction?

j / k navigate · click thread line to collapse

0 comments

7 comments · 2 top-level

somesortofthing1y ago· 5 in thread

reaperman1y ago

We may not see universal adoption in people who are currently >30 but I think we will in the generations that are <25 now.

somesortofthing1y ago

reaperman1y ago

TeMPOraL1y ago

> The fact that this power user class of young people hasn't found any other use cases for LLMs

threeseed1y ago

Of course students are using ChatGPT. It helps them to write assignments.

But it isn't translating into better across the board test results and at least in Australia we would be able to tell because we have yearly standardised testing.

And so schools are looking at it as more of a form of cheating and simply moving back to in-person, hand-written tests.

pas1y ago

with what? how? there's already a lot of bad python/R code out there, how more of it will "overhaul" scientific computing?

> We have essentially standardized a way to approximately solve optimization problems.

.. what does this mean? we had simplex solvers before. do you mean things like protein folding prediction?

j / k navigate · click thread line to collapse