TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

A conversation about AI for science with Jason Pruet

164 点作者 LAsteNERD11 天前

17 条评论

fumeux_fume11 天前
&gt; 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.<p>Most of the bullshitters will tip their hand pretty early that they&#x27;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 never cite a single premise or event to back this up. In the quote above, the phrase &quot;if you&#x27;ve played&quot; stuck out to me. Yes if you play around with them a little it&#x27;s easy to believe they&#x27;re really good at so many things. When you stringently evaluate them, you begin to see they make a lot of mistakes and perform inconsistently on even trivial tasks.
评论 #43969535 未加载
评论 #43969795 未加载
评论 #43970402 未加载
评论 #43974615 未加载
评论 #43972022 未加载
hbartab11 天前
&gt; 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.<p>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 from publicly funded research. What begins as partnership can end in capture.
评论 #43967594 未加载
评论 #43967371 未加载
评论 #43967247 未加载
评论 #43967286 未加载
评论 #43968731 未加载
评论 #43968270 未加载
评论 #43967830 未加载
评论 #43970417 未加载
评论 #43969045 未加载
评论 #43967911 未加载
tantalor11 天前
I was a bit puzzled what &quot;1663&quot; is. Here&#x27;s what I found:<p>&gt; The Lab&#x27;s science and technology digital magazine presents the most significant research initiatives and accomplishments from national-security-related programs as well as projects that advance the frontiers of basic science. Our name is an homage to the Lab&#x27;s historic role in the nation&#x27;s service: During World War II, all that the outside world knew of the top-secret laboratory was the mailing address - P.O. Box 1663, Santa Fe, New Mexico.<p><a href="https:&#x2F;&#x2F;researchlibrary.lanl.gov&#x2F;about-the-library&#x2F;publications&#x2F;" rel="nofollow">https:&#x2F;&#x2F;researchlibrary.lanl.gov&#x2F;about-the-library&#x2F;publicati...</a>
stonogo11 天前
The actual reason is &quot;because they&#x27;re being told to.&quot; Before that, there was a massive public-cloud push DOE-wide. Nobody outside of ASCR is interested in computing, and there&#x27;s a lot of money to be made if you can snag an eternal rent check for hosting federal infrastructure.
senderista11 天前
Clearly AI is worthy of public investment, but given the capture of this administration by tech interests, how can we be sure that public AI funding isn&#x27;t just handouts to the president&#x27;s cronies?
评论 #43967902 未加载
评论 #43967632 未加载
评论 #43968227 未加载
评论 #43968588 未加载
评论 #43967887 未加载
评论 #43968486 未加载
newfocogi11 天前
Another recent AI article out of LANL: <a href="https:&#x2F;&#x2F;www.lanl.gov&#x2F;media&#x2F;publications&#x2F;1663&#x2F;1269-earl-lawrence-ai" rel="nofollow">https:&#x2F;&#x2F;www.lanl.gov&#x2F;media&#x2F;publications&#x2F;1663&#x2F;1269-earl-lawre...</a><p>And discussed on HN: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43765207">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=43765207</a><p>This does feel like a step change in the rate at which modern AI technologies and programs are being pushed out in their PR.
Zorass11 天前
Whether AI is a “good programmer” really depends on what you mean by programming. If it means being fluent in syntax, quickly generating prototypes, and recalling large amounts of code patterns, then yes, it&#x27;s surprisingly strong. But if being a programmer includes debugging intuition, tracking context over multiple sessions, and knowing when not to write code, it&#x27;s still not there yet.
zkmon11 天前
I like how he says that AI is a general-purpose technology like electricity.
评论 #43970446 未加载
andy9911 天前
The real title is &quot;Q&amp;A with Jason Pruet&quot;
lp25111 天前
wonder if they still train all of their models using Mathematica because it was impossible to get pytorch on the classified systems
评论 #43967099 未加载
评论 #43967745 未加载
LAsteNERD11 天前
PR in here for sure, but some smart context on the scientific and nat security potentional the DOE and National Labs see in AI.
zzzeek11 天前
&gt; Over the last two years, we’ve more or less run out of benchmarks where AI isn’t better than humans.<p>this whole &quot;benchmarks&quot; thing is laughable. I&#x27;ve been using Gemini all week to do code assist, review patches, etc. Such impressive text, lists of bullets, suggestions, etc., but then at the same time it makes tons of mistakes, which you then call it on, and it predictably is like &quot;oh sorry! of course!&quot; yes of COURSE. because all it does is guess what word is most likely to come after the previous word. is there a &quot;benchmark&quot; for &quot;doesn&#x27;t hallucinate made up BS?&quot; because humans can do very well on such a benchmark.
quakeguy11 天前
They should invest in natural intelligence first.
paradox46011 天前
Fwiw, LANL saw some of it&#x27;s heaviest layoffs this year, even heavier than those that happened under Nanos in the post Cerro Grande investigation. From what I gather, the feeling up on the hilltop is one of anxiety
KolibriFly11 天前
Feels like we&#x27;re either on the verge of a renaissance or a breakdown, and no one really knows which way it&#x27;ll tilt
bawana11 天前
When will we get AI CEOs? I am tired of seeing corporate decisions driven by quarterly bonuses.
评论 #43968599 未加载
gitroom11 天前
Man, reading all this makes me glad I just write my own scripts when I get stuckAI&#x27;s cool til it isn&#x27;t.