Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A conversation about AI for science with Jason Pruet

lanl.gov

Post reply on HN
Print view View original

121–130 of 163 posts

Re: A conversation about AI for science with Jason Pruet

#121

Post by const_cast »

saretup wrote: post #114
fumeux_fume wrote: post #113

> If you’ve played with the most recent AI tools, you know: They’re very good coders, very good legal analysts, very good first drafters of writing, very good image generators. They’re only going to get better. Most of the bullshitters will tip their hand pretty early that they're just hype men for AI. Right off the bat, the fact that AI is disruptive and transforming society is apparently self-evident because they n…

You sound more prejudiced against AI than they seem to be hyping AI in that sentence. Yes, they’re very good coders; no, they aren’t perfect. Sometimes they make trivial mistakes or hallucinate, and other times they have unique insights about difficult problems or one-shot a lot of complex tasks. If you don’t see the value of AI even at this point, you’re either lacking imagination and/or just too stuck in your own w…

I don't think they're good coders at all, I think they're good at generating small snippets of codes.

If you tried to replace a programmer with whatever LLM it would completely flop. Just write a script to fetch Jira tickets and let it rip - it won't work well. Because it doesn't know that maybe the Jira ticket is stupid, or that if they implement this change it requires insight from Team X, or that this bug is really intended behavior, etc.

Being a programmer is much more than, like, making a neat regular expression or generating slightly customized boilerplate. As it stands, these are assistants in which programmers guide and handhold them the whole way. They are not coders.

Re: A conversation about AI for science with Jason Pruet

#122

Post by groestl »

a_bonobo wrote: post #118
saretup wrote: post #117

Earlier quoted context omitted.

> "they’re very good coders" and "they make trivial mistakes or hallucinate" seems incongruous. It seems that way if you judge them the same way you would human coders, but they’re different. They might be able to do things that veteran coders can’t without spending days on it, and fail at things that beginners can do in half an hour. A car might not be able to traverse difficult terrain as well as a horse, but it do…

Yes, human coders are predictably 'bad' - it scales roughly with the complexity of the task and the years of their experience, along with some randomness ('brain farts'). LLMs are completely unpredictably 'bad' - sometimes a task that a high schooler can solve sends an LLM into a spin, sometimes an LLM solves a task that would require a PhD in the field. The unpredictability is what makes these tools so unfit for pur…

> The unpredictability is what makes these tools so unfit for purpose.

Also, their eagerness to produce. A good employee (senior or not!), knows when they're at the end of their wits and say so. In areas where LLMs are knowledgeable, but not knowledgeable enough, they'll still produce. They're stuck at unconscious incompetence, kind of like a systematic Dunning-Kruger effect. To solve that, we can either increase competence (moving out the area where this effect is noticeable), or maybe introduce some kind of feedback, self reflection (one could say, some kind of conciousness [0]).

[0] inspired by "Goedel, Escher, Bach"

Re: A conversation about AI for science with Jason Pruet

#123

Post by GolfPopper »

bcoates wrote: post #34
candiddevmike wrote: post #28

Earlier quoted context omitted.

How about we fix global warming and switch 100% to clean energy, and then invest in AI?

1. Build atomic power plants sufficient to supply electricity needs for projected future AI megaprojects 2. Inevitable AI winter 3. Keep running the plants, clean energy achieved, stop burning coal, global warming solved

The problem is that's far from a guaranteed path. What if, instead we get:

1. Plan to build atomic power plants sufficient to supply electricity needs for projected future AI megaprojects.

2. Build the AI megaprojects first, because of fallacies related to sunk costs, urgency, and unwillingness to let the pyramid scheme collapse.

3. Never complete the nuclear plants, never get the pot of gold at the end of the AI rainbow, and pump a couple hundred million tons of CO2 into the atmosphere in the mean time.

Re: A conversation about AI for science with Jason Pruet

#124

Post by GolfPopper »

engineer_22 wrote: post #57
candiddevmike wrote: post #28

Earlier quoted context omitted.

How about we fix global warming and switch 100% to clean energy, and then invest in AI?

Let's also cure cancer and stop all wars while we're at it.

You friendly megachurch pastor, I mean TV psychic, I mean AI that will be coming real soon will do that for you! Just send money now.

Re: A conversation about AI for science with Jason Pruet

#125

Post by GolfPopper »

lovich wrote: post #23
cogman10 wrote: post #18

Earlier quoted context omitted.

> I didn't see a reluctance in federal labs to work with capable entrepreneurs My reluctance is when we talk about fraud, waste, and corruptions in government, this is where it happens. The DoD's budget isn't $1T because they are spending $900B on the troops. It's $1T because $900B of that ends up in the hands of the likes of Lockhead martin and Raytheon to build equipment we don't need. I frankly do not trust "entre…

Not that it fully takes away from your argument but a lot of that high price tag is also due to requiring much better controls on material to prevent supply chain attacks ala getting beepers with explosives in the hands of all your leadership

Yet that's the exact opposite of what's been done with something like the F-35[1], with widely distributed production, typically among countries seen as US allies (at least prior to this year), but with key components still made in China.[2] And the problem is even worse in the larger defense industry.[3] Americans pay an immense premium for a military-industrial complex where the PR is largely divorced from reality; for example the USS Gerald R. Ford, commissioned in 2017 still isn't combat ready.[4]

1.https://dsm.forecastinternational.com/2025/04/18/tariffs-tru... 2. https://www.xatakaon.com/materials/u-s-f-35-fighter-jets-and... 3. https://www.forbes.com/sites/davidjeans/2025/04/16/silicon-v... 4. https://www.19fortyfive.com/2025/03/the-navys-ford-class-120...

Re: A conversation about AI for science with Jason Pruet

#126

Post by vanschelven »

keiferski wrote: post #119
fumeux_fume wrote: post #113

> If you’ve played with the most recent AI tools, you know: They’re very good coders, very good legal analysts, very good first drafters of writing, very good image generators. They’re only going to get better. Most of the bullshitters will tip their hand pretty early that they're just hype men for AI. Right off the bat, the fact that AI is disruptive and transforming society is apparently self-evident because they n…

I don’t understand this viewpoint at all. Even if AI tools are nothing more than complex autocomplete systems - and not some kind of new consciousness - that alone is enough to dramatically shift entire industries. And it already is doing so…this isn’t theoretical anymore.

Not OP, but I think the viewpoint here is simply "if someone makes an evidently false claim ('great coders') right off the bat, it's enough to color the reading of the rest of the article. That's independent of what you're saying

Re: A conversation about AI for science with Jason Pruet

#128

Post by KolibriFly »

fumeux_fume wrote: post #113

> If you’ve played with the most recent AI tools, you know: They’re very good coders, very good legal analysts, very good first drafters of writing, very good image generators. They’re only going to get better. Most of the bullshitters will tip their hand pretty early that they're just hype men for AI. Right off the bat, the fact that AI is disruptive and transforming society is apparently self-evident because they n…

The gap between demo impressive and reliably useful in real-world scenarios is still huge in a lot of cases. That said, I think the concern isn't that AI is flawless now, but that it's improving at a pace we haven't really seen with other technologies

Re: A conversation about AI for science with Jason Pruet

#129

Post by keiferski »

vanschelven wrote: post #126
keiferski wrote: post #119

Earlier quoted context omitted.

I don’t understand this viewpoint at all. Even if AI tools are nothing more than complex autocomplete systems - and not some kind of new consciousness - that alone is enough to dramatically shift entire industries. And it already is doing so…this isn’t theoretical anymore.

Not OP, but I think the viewpoint here is simply "if someone makes an evidently false claim ('great coders') right off the bat, it's enough to color the reading of the rest of the article. That's independent of what you're saying

Fair enough, but not sure that's much better of a viewpoint. It might be a worse one. The off-hand comment was "very good coders" (not "great") and dismissing the person's entire thought process because of a nitpick is not a good way to learn anything.

Re: A conversation about AI for science with Jason Pruet

#130

Post by KolibriFly »

hbartab wrote: post #6

> We certainly need to partner with industry. Because they are so far ahead and are making such giant investments, that is the only possible path. And therein lies the risk: research labs may become wholly dependent on companies whose agendas are fundamentally commercial. In exchange for access to compute and frontier models, labs may cede control over data, methods, and IP—letting private firms quietly extract value…

Once public research depends on private infrastructure, the balance of power shifts fast
Post reply on HN