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

Claude 4

anthropic.com

Post reply on HN
Print view View original

811–820 of 1001 posts

Re: Claude 4

#811

Post by runekaagaard »

adamtaylor_13 wrote: post #774
dbingham wrote: post #743

It feels like these new models are no longer making order of magnitude jumps, but are instead into the long tail of incremental improvements. It seems like we might be close to maxing out what the current iteration of LLMs can accomplish and we're into the diminishing returns phase. If that's the case, then I have a bad feeling for the state of our industry. My experience with LLMs is that their code does _not_ cut i…

That’s funny, my experience has been the exact opposite. Claude Code has single-handedly 2-3x my coding productivity. I haven’t even used Claude 4 yet so I’m pretty excited to try it out. But even trusty ol 3.7 is easily helping me out out 2-3x the amount of code I was before. And before anyone asks, yes it’s all peer-reviewed and I read every single line. It’s been an absolute game changer. Also to your point about…

Yeah remember when people were using Claude 3.7... so oldschool man

Re: Claude 4

#812

Post by user_7832 »

Aeolun wrote: post #665
Doohickey-d wrote: post #48

> Users requiring raw chains of thought for advanced prompt engineering can contact sales So it seems like all 3 of the LLM providers are now hiding the CoT - which is a shame, because it helped to see when it was going to go down the wrong track, and allowing to quickly refine the prompt to ensure it didn't. In addition to openAI, Google also just recently started summarizing the CoT, replacing it with an, in my opi…

The Google CoT is so incredibly dumb. I thought my models had been lobotomized until I realized they must be doing some sort of processing on the thing.

You are referring to the new (few days old-ish) CoT right? It’s bizzare as to why google did it, it was very helpful to see where the model was making assumptions or doing something wrong. Now half the time it feels better to just use flash with no thinking mode but ask it to manually “think”.

Re: Claude 4

#813

Post by Crystalin »

hummusFiend wrote: post #299
lelandfe wrote: post #263

Earlier quoted context omitted.

Roomba Terms of Service 27§4.4 - "You agree that the iRobot™ Roomba® may, if it detects that it is vacuuming a terrorist's floor, attempt to drive to the nearest police station."

Is there a source for this? I didn't see anything when Ctrl-F'ing their site.

US Terms of Service 19472§1.117 - "You agree that Google® may, if it detects that it is revealing unconstitutional terms, to hide it instead."

Re: Claude 4

#814

Post by user_7832 »

kovezd wrote: post #595
a2128 wrote: post #580

> Finally, we've introduced thinking summaries for Claude 4 models that use a smaller model to condense lengthy thought processes. This summarization is only needed about 5% of the time—most thought processes are short enough to display in full. Users requiring raw chains of thought for advanced prompt engineering can contact sales about our new Developer Mode to retain full access. I don't want to see a "summary" of…

Don't be so concerned. There's ample evidence that thinking is often disassociated from the output. My take is that this is a user experience improvement, given how little people actually goes on to read the thinking process.

> There's ample evidence that thinking is often disassociated from the output.

What kind of work do use LLMs for? For the semi technical “find flaws in my argument” thing, I find it generally better at not making common or expected fallacies or assumptions.

Re: Claude 4

#815

Post by arewethereyeta »

I feel like these AI companies are in a gold rush while somebody else is selling the shovels. I've never jumped ship for the same service, from a vendor to another... so often. Looks like a race to the bottom where the snake eats itself.

Re: Claude 4

#817

Post by smukherjee19 »

Is there any way to access the models without:

- Linking the chats with my personal account - Having Anthropic train the model with my data?

Like, having the knowledge of the model with the privacy of local LLMs?

Re: Claude 4

#818

Post by jakemanger »

Been playing around with it in Cursor and have to say I'm pretty dang impressed.

Did notice a few times that it got stuck in a loop of trying to repeatedly make its implementation better. I suppose that is ok for some use cases but it started overthinking. I then gently prompted it by saying "you're way overthinking this. Just do a simple change like ..."

I guess there's still a purpose for developers

Re: Claude 4

#819

Post by noduerme »

csomar wrote: post #805
noduerme wrote: post #763

Earlier quoted context omitted.

Wouldn't that be the best thing possible for our industry? Watching the bandwagoners and "vibe coders" get destroyed and come begging for actual thinking talent would be delicious. I think the bets are equal on whether later LLMs can unfuck current LLM code to the degree that no one needs to be re-hired... but my bet is on your side, that bad code collapses under its own weight. As does bad management in thrall to tr…

That's not how real life works; you are thinking of a movie. Management will never let down of any power they accumulated until the place is completely ransacked. The Soviet Union is a cautionary tale, a relatively modern event and well documented.

I only work for companies where I have direct interaction with the owners. But I think that any business structure that begins to resemble a "soviet" type, where middle management accumulates all the power (and is scared of workers who have ideas) is inevitably going to collapse. If the way they try in the late 2020s to accumulate power is by replacing thoughtful coders with LLMs, they will collapse in a very dramatic, even catastrophic fashion. Which will be very funny to me. And it will result in their replacement, and the reinstatement of thoughtful code design.

A lot of garbage will have to be rewritten and a lot of poorly implemented logic re-thought. Again, I think a hard-learned lesson is in order, and it will be a great thing for our industry.

Re: Claude 4

#820

Post by jmcgough »

smukherjee19 wrote: post #817

Is there any way to access the models without: - Linking the chats with my personal account - Having Anthropic train the model with my data? Like, having the knowledge of the model with the privacy of local LLMs?

Per their data privacy policy, they do not train models on customer conversations by default.
Post reply on HN