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

The Whole Anthropic Kerfuffle

twitter.com

Post reply on HN
Print view View original

61–70 of 89 posts

Re: The Whole Anthropic Kerfuffle

#61

Post by sfifs »

adampunk wrote: post #7
jameskilton wrote: post #2

> At Anthropic, we build AI to serve humanity’s long-term well-being. If Anthropic actually cared about humans, they would have the best customer support (staffed by humans, for humans) and communications team (again, staffed by humans, for humans). As both of these are actually on par with Silicon Valley standards (between medicore and atrociously bad), Anthropic cannot and should not be trusted with anything to do…

How would you staff a support line for a product with a billion users?

With lower margins of course. Walmart, Indian Railways, major airlines etc all support massive user bases comparable to or bigger than the paid tiers of these apps. But of course the cult of Big Shareholder value creation means the CEO that does this, especially in the US will be fired.

Re: The Whole Anthropic Kerfuffle

#62

Post by sfifs »

adampunk wrote: post #48
0gs wrote: post #13

Earlier quoted context omitted.

it's hard, but not THAT hard, to find a few dozen people who can deal with large volumes of support tickets every day. so for a company like anthropic, you'd use a customized claude to triage and then those few dozen people spend all day actually caring about solving users' problems. a contract with fin fka intercom (lol) to offload this is a step in the wrong direction imo, but then nobody pays for support so it's h…

I'm sorry but a few dozen people actually caring about the problems of a billion users is a fart in a windstorm. You might as well hire a half-dozen to care, or none, for all the work you'll do. You'd need a dozen people just to design a scheduler for handling tickets only to watch that catch fire too. I don't get it. None of the hyperscalers have human support teams at scale because it's obviously infeasible. Why, j…

Not infeasible, just allows lower net margins.

Re: The Whole Anthropic Kerfuffle

#63

Post by ochronus »

quietsegfault wrote: post #60
ochronus wrote: post #43

Yeah, claude code is mostly unusable at this point. It's been in a constant decline for a good 1-2 months. It's more like a scam now.

What are the biggest problems you've seen? Is it mostly related to limits for the subscription plans? Within my circles (mostly big enterprises), I see more and more of my friends using Claude, and spending money on it, so they must be getting some sort of value out of it. For my uses, I've also been successful with Claude Code, though someone else is paying for my tokens.

Yes, it's related to the limits of the subscription plans. 2ish months ago the same sub started hitting limits earlier and earlier on very similar tasks/codebases. Then they said, Oh, sorry, it was a caching bug, now it's fixed. It wasn't. Then a couple more bugs. Still no fix. Recently they announced "doubling the limits" (made possible by the Grok deal) - still hitting limits super fast compared to how it used to be 2+ months ago. Moreover, the models got somewhat dumber and slower, too.

Re: The Whole Anthropic Kerfuffle

#64

Post by pessimizer »

XorNot wrote: post #42
dormento wrote: post #19

Earlier quoted context omitted.

That's true. But keep in mind capital can be trivially used to influence policy so that local ones are eventually disallowed.

When has that ever happened though? LLMs are software there's no plausible way to stop them running locally.

They're already nearly there. Once everybody's computers are rooted for age verification, with government-approved OSes, it's barely even a step to only allow government-approved AI on them.

For some reason, people assume that mandatory age/identity verification on people's machines is primarily geared toward reporting across the network. If you wanted to age-restrict across the network, there are any number of trivial ways to do it that they have never even entertained. The reason that the first victims of this legislation are operating systems is because they, like Apple, Google, and Microsoft already have, want to restrict the software you can run.

They have to make sure that you can't have AI that can make pictures of naked children, or naked anything, or that might "underreport" the number of casualties in the Tienanmen Square protests, or could cast doubt on the vaccine (whichever vaccine), or say anything that could be interpreted as anti-Semitic (like that the Palestinians aren't savage animals), etc., etc....

This is already easy to get public support on, because in the same way they whipped up bizarre mind control allegations against genuinely evil social media companies to throw the public off the scent, the public is being groomed with absolutely bizarre and incoherent predictions of the evils of generative AI in order to throw them off the actual evils of the people behind generative AI. The same way that the anti-social media agitprop just resulted in TikTok being sold to explicit propagandists during a genocide and age attestation (as the social media giants do business without interruption), the "AI scaremongering" is just going to result in physical restrictions on individuals running AI - it will be tracked like explosives or nuclear material. The giant AI companies will be sold as the solution, just like closed platform software "stores" from Apple and Google are sold as consumer protection.

> When has that ever happened though?

Microsoft has to sign Linux so it can be installed under Secure Boot. Encryption was regulated as arms export, and is fully under attack again.

Re: The Whole Anthropic Kerfuffle

#65

Post by parliament32 »

infecto wrote: post #55
hdndjsbbs wrote: post #46

Earlier quoted context omitted.

They have to speedrun boiling the frog because the capital expenditure is insane. Remains to be seen just how fast you can boil a frog before the frog notices

Disagree. Most businesses of size are going to enterprise agreements which are all on demand rates. Those rates have not been changing other than the underlying cost to the model API rates fluctuating. You could make argument they are secretly using that has the price lever. With volume enterprises can already negotiate lower token rates. I don’t see a boiling the frog situation.

They will still need to increase costs for enterprise to be profitable, they're just going to be more greasy about it. Claude 5 will cost 20% more but not be 20% better, more shenanigans with "oh no we had a bug in our cache system :^)", or this gem from the current enterprise pricing page: "Opus 4.7 uses a new tokenizer... may use up to 35% more tokens for the same fixed text".

Re: The Whole Anthropic Kerfuffle

#66

Post by electric_mayhem »

ochronus wrote: post #63
quietsegfault wrote: post #60

Earlier quoted context omitted.

What are the biggest problems you've seen? Is it mostly related to limits for the subscription plans? Within my circles (mostly big enterprises), I see more and more of my friends using Claude, and spending money on it, so they must be getting some sort of value out of it. For my uses, I've also been successful with Claude Code, though someone else is paying for my tokens.

Yes, it's related to the limits of the subscription plans. 2ish months ago the same sub started hitting limits earlier and earlier on very similar tasks/codebases. Then they said, Oh, sorry, it was a caching bug, now it's fixed. It wasn't. Then a couple more bugs. Still no fix. Recently they announced "doubling the limits" (made possible by the Grok deal) - still hitting limits super fast compared to how it used to b…

Also, they seem to ignore auto mode, seizing on any reason to stop, and sometimes just stopping despite saying they’re proceeding with something.

Basically, ongoing enshittification.

Re: The Whole Anthropic Kerfuffle

#67

Post by ochronus »

nickdothutton wrote: post #52

I really wish Anthropic would consult some monetization experts. Their recent strategy has been all over the place and they are burning early goodwill.

I don't think they are clueless, but rather, struggling. Being an AI provider is a money burner, and they probably don't have enough fuel for the fire anymore, so they are trying things to squeeze more $$$ and limit usage at the same time.

Re: The Whole Anthropic Kerfuffle

#68

Post by gruez »

parliament32 wrote: post #54
AlexandrB wrote: post #45

Earlier quoted context omitted.

How do we tell the doordash/uber playbook from the moviepass playbook? Because the latter would be awful to build your business on.

Moviepass (afaik) was an attempt at the exact same playbook, it just failed. Anthropic will also fail when the competition is.. near-equivalent-capability DeepSeek/Qwen/Llama on a $1k GPU with a break-even of 5 months of subscription costs. The value is simply not there for what they would need to charge to become profitable.

>when the competition is.. near-equivalent-capability DeepSeek/Qwen/Llama on a $1k GPU with a break-even of 5 months of subscription costs

Lol no. Chinese AIs are definitely not "near-equivalent-capability". The empirical proof is pretty obvious: how many people have you heard talking about using their codex/claude code subscription vs their z.ai or qwen subscription? Moreover even the Chinese models require epic amounts of GPUs to run the full version, eg. https://apxml.com/models/glm-51 needs 1515 GB to run, and that's with a measly 1024 token context. To get it to run on your "$1k GPU" you'd need to quantize it, making it even dumber.

Re: The Whole Anthropic Kerfuffle

#69

Post by parliament32 »

afavour wrote: post #36
parliament32 wrote: post #18

It really is the doordash/uber playbook all over again eh? Sell at a massive loss, gain userbase, then gradually boil the frog by adding fees, removing features, and increasing prices. Except instead of doing this a few years down the line, they're speedrunning the tighten-the-noose phase. Unfortunately the competition is nipping at their heels so there's a good chance this blows up in their faces.

I still think/hope/pray the future will be on-device models that don't need constant retraining. That will blow up the existing business model but I think a company could still make good money with a "majority local/remote for the really challenging stuff" model. The problem is that today's AI companies have taken on so much funding that a reasonable, not crazy profit ratio isn't enough for them.

The future is already pretty much here. Note the recent stories about Chrome adding a local model, not to mention the Googlebook demo (if it works as advertised, there's a 0% chance you could get that kind of latency with a non-local model).

Re: The Whole Anthropic Kerfuffle

#70

Post by parliament32 »

gruez wrote: post #68
parliament32 wrote: post #54

Earlier quoted context omitted.

Moviepass (afaik) was an attempt at the exact same playbook, it just failed. Anthropic will also fail when the competition is.. near-equivalent-capability DeepSeek/Qwen/Llama on a $1k GPU with a break-even of 5 months of subscription costs. The value is simply not there for what they would need to charge to become profitable.

>when the competition is.. near-equivalent-capability DeepSeek/Qwen/Llama on a $1k GPU with a break-even of 5 months of subscription costs Lol no. Chinese AIs are definitely not "near-equivalent-capability". The empirical proof is pretty obvious: how many people have you heard talking about using their codex/claude code subscription vs their z.ai or qwen subscription? Moreover even the Chinese models require epic amo…

Today, sure. But we already see diminishing returns with Claude releases, and we know the open models are closing the gap (~6 months behind according to the benchmarks). And when the pitch is "our models are 5% better but cost $200/mo.. also here's a mountain of restrictions" it just won't make sense anymore. Give it a year or two.

I could see the "avoid the hardship of running a local model for $20/mo" angle but Anthropic has shown they have little interest in those customers.

Post reply on HN