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

Switch to Claude without starting over

claude.com

Post reply on HN
Print view View original

251–260 of 287 posts

Re: Switch to Claude without starting over

#251

Post by kgeist »

Wowfunhappy wrote: post #142

I don't understand how people use these apps with memory enabled. I am always carefully controlling the context of each conversation. The idea that past conversations could bleed into current ones is unthinkably terrible.

Yeah, I turned it off completely after it once started generating random garbage in long conversations with memory enabled. Like, it would give completely unrelated replies: I'd ask about Go, and it would tell me about the weather in Paris (or something else we discussed once months ago). Maybe it was just a temporary bug, but it was an eye-opener that context poisoning exists.

Re: Switch to Claude without starting over

#252

Post by wiredpancake »

0_____0 wrote: post #178
deaux wrote: post #68

Earlier quoted context omitted.

The reply about knowledge about their job and familt made me think. The only thing I can now think of is using it as a personal therapist. Or asking how to approach their kids. And they're a bit embarrassed about it, because it's still outside the Overton window -especially on HN - which is why they aren't sharing it. If someone has different usecases, please do prove me wrong! Maybe I just lack imagination.

Such an incredible amount of personal, intimate knowledge to share with a company. Sure, Google can figure out where I live and who I visit because I have an Android phone, but they'll never know the contents of those relationships. I have a line in the sand with the AI vendors. It's a work relationship. If I wouldn't share it with a colleague I didn't know super well, I'm not telling it to a AI vendor.

[dead]

Re: Switch to Claude without starting over

#253

Post by davidee »

kdheiwns wrote: post #92
Joeri wrote: post #13

I already switched to claude a while ago. Didn’t bring along any context, just switched subscriptions, walked away from chatgpt and haven’t touched it again. Turned out to be a non-event, there really is no moat. I switched not because I thought Claude was better at doing the things I want. I switched because I have come to believe OpenAI are a bad actor and I do not want to support them in any way. I’m pretty sure t…

Yesterday was my first time trying it. One thing that felt a bit strange to me was that I asked it something and the response was just one paragraph. Which isn't bad or anything but it felt... strange? Like I always need to preface ChatGPT/gemini/whatever question with "Briefly, what is..." or it gives me enough fluff to fill a 5 page high school essay. But I didn't need to do that and just got an answer that was to…

That tracks for me; longtime claude, claude code pro subscriber (not all of it has been good - but that's neither here nor there).

Over the last few iterations of Sonnet and Opus, anthropic has definitely trained me to ask it to explain something "in detail" (or even "in great detail") when I want as much nuance as possible.

It used to be the inverse - way too much detail when I didn't want it.

Re: Switch to Claude without starting over

#254

Post by ivan_gammel »

re-thc wrote: post #136
ivan_gammel wrote: post #99

Earlier quoted context omitted.

The more code is out there, the worse is the average in the training dataset. There will be legacy approaches and APIs, poor design choices, popular use cases irrelevant for your context etc that increase the chances of output not matching your expectations. In Java world this is exactly how it works. I need 3-5 iterations with Claude to get things done the way I expect, sometimes jumping straight to manual refactori…

> To overcome this problem model needs specialized training, that I don‘t think the industry knows how to approach We already have coding tuned models i.e. Codex. We should just have language / technology specific models with a focus on recent / modern usage. Problem with something like Java is too old -- too many variants. Make a cut off like at least above Java 8 or 17.

> We should just have language / technology specific models with a focus on recent / modern usage.

The “just” part is a big assumption. It is far from easy, given that modern best practices are always underspecified. The effective model for coding must have reasoning signals to be much stronger than coding patterns, and that, I suspect, requires very different architecture.

Re: Switch to Claude without starting over

#256

Post by toss1 »

thunky wrote: post #239
toss1 wrote: post #172

Earlier quoted context omitted.

If each of us individually or as corporations should not be in the business of deciding what it "evil", who should be in that business? Everyone SHOULD continuously consider, decide, and live by moral judgements and codes they internalize, and use to make choices in life. This aspect of life should NEVER be outsourced — of course, learn from and use codes others have developed and lived by — but ALWAYS consider deepl…

> If each of us individually or as corporations should not be in the business of deciding what it "evil", who should be in that business? This is easy imo. Two methods: 1. The law. It should not be legal for the US Govt to murder people at will. If it is legal, then of course they'll use tools to make it easier. Maybe AI, maybe Clippy. If they can't use AI then they'll fall back to using some other way of doing it li…

Exactly zero of those account for an individual's or company's ability tolive by their own moral code

And this AI software is not a mere static object like a hammer that can be handed off to a customer and what it is used for is their business, to build a house or bash a living skull.

This is a system that must be constantly maintained by it's builders.

Moreover, even if we use your standard, the law, it has already been decided in Anthropic's favor.

What you require is that Anthropic actively participate in activities that they consider abhorrent and/or unwise. SCOTUS has already ruled that a business cannot even be required to sell a cake to someone if it does not like the intended purpose (in that case, at a celebration at a gay wedding).

Re: Switch to Claude without starting over

#257

Post by pants2 »

neya wrote: post #210
pants2 wrote: post #189

Earlier quoted context omitted.

OpenAI actually does have two excellent OSS models. Not Anthropic. Not that OpenAI is 'open' per se, but more so than Anthropic. Also see the Codex vs Claude Code extensibility.

They are far from excellent and they were open sourced due to the mounting pressure for calling themselves "Open" AI and not doing anything open. At the time, they also had Chinese competitors wiping the market value of many stocks (NVidia, etc.) after releasing true OSS models that performed as good as SOTA models and they had to retaliate. I don't know of anyone who uses those OSS models in production instead of Qw…

I use them heavily in production because OSS 120B on Groq/Cerebras is one of the fastest models available with enough smarts for basic tasks

Re: Switch to Claude without starting over

#258

Post by idopmstuff »

wps wrote: post #2

Could someone explain the appeal of account-wide memory to me? Anthropic’s marketing indicates that nothing bleeds over, but I’m just so protective of my context that I cannot imagine having even a majorly distilled version of my other chats and preferences having on weight on the output. As for certain preferences like code styling or response length, these are all fit for custom instructions, with more detailed thi…

I use Claude code in a number of different parts of my business - coding internal applications, acting as a direct interface to SaaS via APIs and just general internal use.

I find there is a virtuous cycle here where the more I use it, the more helpful it is. I fired my bookkeeper and have been using Claude with a QBO API key instead, and because it already had that context (along with other related business context), when I gave it the tax docs I gave to my CPA for 2024's taxes plus my return, and asked it to find mistakes, it determined that he did not depreciate goodwill from an acquisition. CPA confirmed this was his error and is amending my return.

Then I thought it'd be fun to see how it would do on constructing my 2024 return just from the same source docs my CPA had. First time I did it, it worked for an hour then said it had generated the return, checked it against the 2024 numbers and found they're the same. I had removed the 2024 before having it do this to avoid poisoning the context with the answers, but it turns out it had a worksheet .md file that it was using on prior questions that I had not erased (and then it admitted that it had started from the correct numbers).

In order to make sure I wouldn't have that issue again, I tried the 2024 return again, completely devoid of any historical context in a folder totally outside of my usual Claude Code folder tree. It actually got my return almost entirely correct, but it missed the very same deduction that it had caught my CPA missing earlier.

So for me, the buildup of context over time is fantastic and really leads to better results.

Re: Switch to Claude without starting over

#259

Post by gizmodo59 »

It’s disgusting how they have successfully fooled people into thinking they are the good guys. They partnered with palantir, let them freely do the dirty work and once they realized they can make money directly they spin the PR and just trying to get more users. Well played.

I wish oss models are good so that we don’t have to deal with either leading companies!

Re: Switch to Claude without starting over

#260

Post by thunky »

toss1 wrote: post #256
thunky wrote: post #239

Earlier quoted context omitted.

> If each of us individually or as corporations should not be in the business of deciding what it "evil", who should be in that business? This is easy imo. Two methods: 1. The law. It should not be legal for the US Govt to murder people at will. If it is legal, then of course they'll use tools to make it easier. Maybe AI, maybe Clippy. If they can't use AI then they'll fall back to using some other way of doing it li…

Exactly zero of those account for an individual's or company's ability tolive by their own moral code And this AI software is not a mere static object like a hammer that can be handed off to a customer and what it is used for is their business, to build a house or bash a living skull. This is a system that must be constantly maintained by it's builders. Moreover, even if we use your standard, the law, it has already…

> even if we use your standard, the law, it has already been decided in Anthropic's favor.

I support Anthropic here. They had a deal with the Govt and the Govt bullied them. That should not be allowed, and Anthropic is suing which makes sense to me. Anthropic should be allowed to set any terms of use for the product that they want, and gain or lose business based on those terms. That's fine.

I'm saying that the failure is actually upstream. It should not be possible for Anthropic's AI to be used to mass surveile or murder people, because those things should be illegal by law and the govt should not be allowed to do it and should not be doing it. Somehow it isn't this way though.

So now that we find ourselves in this failed state, we have to rely on Anthropic to be "the law": to identify what what's "evil" and disallow it. I'm saying that's out of scope for a tech company and they shouldn't be expected to do that. They should only be in the business of making good tech and then be free to let it be used by anyone for any purpose that that the law allows.

This also means that if it's illegal to share information on how to build a bomb without AI, then it should be illegal for Claude to share that information with AI. So Anthropic to does need to make sure they're not breaking the law themselves as well.

Post reply on HN