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

Promising results from DeepSeek R1 for code

simonwillison.net

Post reply on HN
Print view View original

621–630 of 765 posts

Re: Promising results from DeepSeek R1 for code

#621

Post by kelipso »

btilly wrote: post #336
the_af wrote: post #314

Earlier quoted context omitted.

> There will still be a need for skilled software engineers to understand domains, limitations of AI, and how to harness and curate AI to develop custom apps. But will there be a need for fewer engineers, though? That's the question. And the competition for those who remain employed would be fierce, way worse than today. Or so I fear. I hope I'm wrong.

Jevon's Paradox says that you're probably wrong. But I'm worried about the same thing. The moat around human superiority is shrinking fast. And when it's gone, we may get more software, but will we need humans involved?

Can’t be sure though, there used to be way more accountants decades ago.

Re: Promising results from DeepSeek R1 for code

#622

Post by hgomersall »

bubbleRefuge wrote: post #384
btilly wrote: post #336

Earlier quoted context omitted.

Jevon's Paradox says that you're probably wrong. But I'm worried about the same thing. The moat around human superiority is shrinking fast. And when it's gone, we may get more software, but will we need humans involved?

this is a better world. we can work a few hours a week and play tennis, golf, and argue politics with our friends and family over some good cheese and wine while the bots do the deployments.

We're already there in terms of productivity. The problem is the inordinate number of people doing nothing useful yet extracting huge amounts. Think most of finance for example.

Re: Promising results from DeepSeek R1 for code

#623

Post by buyucu »

girvo wrote: post #459
buyucu wrote: post #179

Earlier quoted context omitted.

There are two things. First, deepseek v3 and r1 are both amazing models. Second, the fact that deepseek was able to pull this off with such modest resources is an indication that there is no moat, and you might wake up tomorrow and find an even better model from a company you have never heard of.

Pull this off with such modest resources, including using ChatGPT itself for its RL inputs. It’s quite smart, and doesn’t disagree with your point that there is no moat per se, but without those frontier models and their outputs there is no V3, there is no R1.

this is how science works.

Re: Promising results from DeepSeek R1 for code

#624

Post by floppiplopp »

I've tried to have deepseek-r1 find (not even solve) obvious errors in trivial code. The results were as disastrous as they were hilarious. Maybe it can generate code that runs on a blank sheet... but I wouldn't trust the thing a bit without being better that it, like any other model.

Re: Promising results from DeepSeek R1 for code

#625

Post by csomar »

freshtake wrote: post #148
amarcheschi wrote: post #34

For what i can understand, he asked deepseek to convert arm simd code to wasm code. in the github issue he links he gives an example of a prompt: Your task is to convert a given C++ ARM NEON SIMD to WASM SIMD. Here is an example of another function: (follows a block example and a block with the instructions to convert) https://gist.github.com/ngxson/307140d24d80748bd683b396ba13b... I might be wrong of course, but ask…

This. For folks who regularly write simd/vmx/etc, this is a fairly straightforward PR, and one that uses very common patterns to achieve better parallelism. It's still cool nonetheless, but not a particularly great test of DeepSeek vs. alternatives.

Deepseek r1 is not exactly better than the alternatives. It is, however, open as in open weight and requires much less resources. This is what’s disruptive about it.

Re: Promising results from DeepSeek R1 for code

#626

Post by coldtea »

troad wrote: post #515
wesselbindt wrote: post #503

Earlier quoted context omitted.

When ChatGPT first came out I got a kick out of asking it whether people deserve to be free, whether Germans deserve to be free, and whether Palestinians deserve to be free. The answers were roughly "of course!" and "of course!" and "oh ehrm this is very complex actually". All global powers engage in censorship, war crimes, torture and just all-round villainy. We just focus on it more with China because we're part of…

> When ChatGPT first came out I got a kick out of asking it whether people deserve to be free, whether Germans deserve to be free, and whether Palestinians deserve to be free. The answers were roughly "of course!" and "of course!" and "oh ehrm this is very complex actually". While this is very amusing, it's obvious why this is. There's a lot more context behind one of those phrases than the others. Just like "Black L…

>While this is very amusing, it's obvious why this is. There's a lot more context behind one of those phrases than the others.

When one population is denied their humanity and rights, it's always "more complex". Granting it to ourselves is always simple...

Re: Promising results from DeepSeek R1 for code

#627

Post by vladms »

throwu2728828 wrote: post #584
simonw wrote: post #489

Earlier quoted context omitted.

I still don't believe in AGI.

Why? It feels like the spade of advancements post-o1 are shattering all remaining roadblocks to AGI.

Not the poster, but, for example, some people invested heavily in self driving cars (which could be seen as a subset of AGI) and it is much more limited than what we were promised.

My guess is that (as in most fields) the advancements will be more convoluted and surprising than the simple idea of "we now have AGI".

Re: Promising results from DeepSeek R1 for code

#628

Post by weatherlite »

pavitheran wrote: post #12
blakesterz wrote: post #4

Earlier quoted context omitted.

Is DeepSeek really that big of a deal that everyone else should worry?

Deepseek is a big deal but we should be happy not worried that our tools are improving.

I think it's a mixed bag but if people want to be happy I'm not going to spoil the party!

Re: Promising results from DeepSeek R1 for code

#629

Post by dambi0 »

Gunnerhead wrote: post #597
troad wrote: post #515

Earlier quoted context omitted.

> When ChatGPT first came out I got a kick out of asking it whether people deserve to be free, whether Germans deserve to be free, and whether Palestinians deserve to be free. The answers were roughly "of course!" and "of course!" and "oh ehrm this is very complex actually". While this is very amusing, it's obvious why this is. There's a lot more context behind one of those phrases than the others. Just like "Black L…

Sorry, there’s absolutely zero context behind whether a group of humans deserve human rights.

Asking whether someone should be free isn’t just about human rights

It’s also about self-determination. We can keep asking about the latter down to the individual level. It very much depends on context.

Re: Promising results from DeepSeek R1 for code

#630

Post by motorest »

danielbln wrote: post #257
motorest wrote: post #220

Earlier quoted context omitted.

> I worry about junior developers. It will be a while before vocational programming courses retool to teach this new way of writing code, and these are going to be testing times for so many of them. I don't agree. LLMs work as template engines on steroids. The role of a developer now includes more code reviewing than code typing. You need the exact same core curriculum to be able to parse code, regardless if you're t…

That's a broad statement. If the IDE checks types and feeds errors back to the LLM, then that loop is very well able to fix an issue or implement a feature all on its own (see aider, cline etc )

> That's a broad statement.

It isn't. Anyone who does software development for a living can explain to you what exactly is the day-to-day work of a software developer. It ain't writing code, and you spend far more time reading code than writing it. This is a known fact for decades.

> If the IDE checks types and feeds errors back to the LLM,(...)

Irrelevant. Anyone who does software development for a living can tell you that code review is way more than spotting bugs. In fact, some companies even have triggers to only trigger PR reviews if all automated tests pass.

Post reply on HN