simianwords

> I don't have a problem with experimenting on a mouse if there's a reasonable chance the end result will save lives or meaningfully help people, and precautions are taken to prevent suffering.

> so lets assume some minute expected benefit.

Crucial part: all of humanity enjoys the benefit so you must multiply it by the population. So lets assume the expected benefit after adjusting for uncertainty is around 10^-6. And assume the benefit of eating a chicken is around 10 units.

Lets assume the pain felt by the mouse is around 2 pain units and the chicken's pain is around 2 as well.

Assume the benefit you get for generating 2 pain units after eating a chicken is 10 units.

Whereas the benefit from the mouse experiment after generating 2 pain units is 10^-6 * 10^9 = 1000.

Lets also put a subjective factor: eating meat is mostly hedonistic and testing on animals doesn't really have alternatives. I'm being honest here, if AI can eventually do surgery by itself then I'd like it to be tested on animals first and I'd prefer people don't protest against this especially those people who already eat meat.

Also lets not forget the extent of animals being killed for meat: its around 100 billion animals a year for mostly hedonistic reasons that can be fully avoided. Compared to that, animals being killed for medical research is around 100million. That's around 3 OOM less than for meat.

So yes, I think it is a complete non issue that a single mouse is killed for non hedonistic reasons and for benefit of science and I think a person has lost sight of reality if they think it is comparable to eating meat.

I'm preempting a reply like "haha you just made those numbers up so I won't respond". Ok you can do that. But this has been my point from the start - the benefits of animal research diffuse to all of humanity whereas eating meat is a purely selfish and hedonistic experience.

After reading this story I understood that a lot of pieces had to come together to make the plan work. It was a huge long shot from the start and that makes me wonder how many times they must have tried this in the past? To make long shots like this work they must have tried relentlessly; they would've tried hundreds of thousands of developers before this

Honest question, I wonder what postman provides now that I do all API shaped thing in Codex. I can't imagine myself going back to hand-writing json's in a UI.

Not to be discouraging but I can bet 100:1 that Fable with the correct prompt can beat any fine tuned model. Unless you are seriously optimising for cost, then I'd put it 5:1.

I'm a bit uncomfortable with the "think of the children" kind of argument.

I don't think stories are evil because it is deleted. That means disappearing texts are also evil in Signal. Just because they are kids doesn't make disappearing content suddenly evil.

Sharing of location is also not evil and it is egregious to invoke a pathological case of a kid sharing a location with a predator to justify the arguments.

If these were really concerns then all these things should be severely restricted for children

1. disappearing messages on whatsapp

2. hell whatsapp itself because it can be used to talk to predators

3. almost all social media like Reddit, X

4. online gaming because it can foster bullying and also predators

I find that AI should be used more extensively for System Design. I have seen a lot of people not use AI for that and come up with frighteningly complex designs and let their agents run loose to implement it.

Retroactively speaking, I engaged with the already agreed designs and used AI to find much better ways to do the same thing.

Overall, I wish people used AI even for System Design as that seems to have a higher impact.

I responded completely reasonably that the cost amortises. You instead brought up an absurd scenario that there’s an equal chance that someone else gets killed.

That’s a non sequitur because any action whatsoever has that property - you going outside your house may also result in a meteorite falling, would I ask you not to do it?

The mouse experiment is different because it was done for the intention of saving future lives. Not that intention matters but the everything around it implies that this mouse dying will have epsilon chance of helping humanity unlike other tasks which all else equal will not help anyone in surgery.

I’ll let it for anyone reading this to decide how to interpret this thread. And also how to judge a persons virtue if they care this much about a single mouse and not billions of chickens, pigs, cows dead every year for just purely hedonistic reasons.

I wish fewer animals were killed for meat and I wish fewer people took a * single * mouse’s death as an opportunity to show their “virtue”

I love the kind of person who eats meat everyday but god forbid a single mouse is used for an experiment!

It’s just a way to show one’s virtue

Also don’t go outside your house because you may stamp on ants smh

> The manager always has a very good stick: “Do a good job or you will be fired”. And me wanting food is well aligned to not produce buggy code. Or make my job harder in the future by writing unmaintainable code.

You are welcome to not use the model if it makes mistakes. Its weird that you are cooking up some edge case scenario to make your argument strong - everyone is incentivised to make things work well. Anthropic wants their models to perform well because that's why you will buy it.

> Someone needs to be aware of how it’s processed, so that when malformed data appears

Eh no. You need to know about only a compressed part of it, not 100%. Do you really claim that you need to know 100% of the code? This is not true today and certainly won't be true in the future. Hardly any code will be looked at by humans in the same way my manager doesn't look at my code.

Lets agree that there's like 0.000001% chance that it helped. It amortises over whole of humanity itself unlike you eating meat.

I don't care for mice and the small harm that it did do will be amortised over all the gains. Unlike eating meat which keeps repeating.

no, it is completely illogical. Animal testing at least has amortisation properties and is utilitarian at its core. Factory farming and eating animals for food can be completely avoided.

And that some one can be an LLM. Does your manager at your job review every line of code you wrote or trust you based on your competence and reviews only a compressed portion?

Frankly, these discussions are tiring because there's a false notion that humans have to be in the loop 100% of the time. This is untrue. LLMs are accurate enough that they can be left to do things on their own and human intervention is needed only for out of distribution instances.

I'm not sure why this is downvoted. This is the correct answer. Code is mostly written by and for agents with the option for humans to read it from time to time.

This kind of fallacy keeps repeating time and time again. Another version of this fallacy goes like this: "agents can create a lot of code but humans have to review each PR anyway so reviewing is a massive bottleneck". This doesn't make sense because reviewing is also done by agents and humans don't have to read every line.

There's unfortunately a lot of emotions and cope around this so you will be downvoted. There was a recent viral news article implying that Anthropic had lost 80B in a single quarter, this was really because of increased valuations. But a lot still believe this is the case and Anthropic will continue to have this 80B loss..

The reality is that they are profitable in the normal sense

Unironically you should be using claude for monitoring and within CI. "PRs causing bottleneck" is an anti pattern during AI. It is as easy to verify as it is to generate.

This is naive. AI is one of those few things that does have moat. Its shockingly hard to hire people and create a good company culture that does frontier training. There's only like 3-4 labs in the world and the rest are distilling.

Even in next 3-4 years, the company that has the best ecosystem of researchers, culture, shared knowledge and processes will be able to churn out models in a way that can't be replicated.

New bubble pops theory found: the 15th derivative is negative guys! Ed Zitron's 10000 word slop essay to follow

I love coming to HN because I get to see comparisons like this between Ed Zitron and Dario.

I find Dario interesting, serious and competent.

Dario has been talking about AI safety for years, even at 2015. So its not like some grift - that's what he really truly believes and he's putting where his mouth is and has been able to create what would be the highest market cap company in the world.

eh sometimes you don't need any thorough dismissal. for all the "AI does make you productive" headlines, there's bound to be a few that goes the opposite way. Its just statistics, and these headlines get upvoted because of emotional reasons.

There's no way AI is going away, nor any way companies are going to use AI even less than they are today. Hell, even Linus Torvalds is using AI. At this point, there's no need for a serious rebuttal to these arguments.

We have Linus Torvalds himself using and promoting AI. This kind of rhetoric is ridiculous. It feels like coping

new bubble pops theory found - random AI (not even sure which model) can't spell seventeen.

FWIW I tried to reproduce in ChatGPT. It works well.

It’s not clear what your argument is.

It’s more or less certain that Anthropic is already profitable if you discount the losses due to becoming too valuable.

What you are saying here

> The CapEx spend for AI not only makes AI itself more expensive (and thus less profitable),

There’s nothing inherently unique about this. It’s just supply and demand - the only problem is that the ecosystem of raw materials can’t keep up with the demand. It will catch up soon and the prices will stabilise.

Your only problem seems to be this aspect only: that RAM production hasn’t increased according to demand.

No I don’t agree they should slow down. I think RAM and other things should rather speed up to react appropriately to the demand.

It’s like saying the iPhone is not profitable because too many people want it and its price is going up.

I tried to give this article an honest read but

> Last year, for instance, Anthropic lost $80 billion, and Amazon had negative $42 billion free cash flow. The entire big tech complex, it seems, has moved from minting cash to burning it. And investors are, well, not panicking exactly, but wondering if the bubble is ending

Is this really the quality of such newsletters? I’m not asking rhetorically because I’d like to also have trustable sources to understand the underlying reality. I actually spent time understanding this part and the 80B dollar loss thing is completely misrepresented here (also went viral in other media).

I think this is incorrect logic. OpenAI’s valuation doesn’t come from the sum of other tech company’s valuation.

You do know that all valuations can go up together right? It’s not zero sum. OpenAI can create trillions in value not by taking it from somewhere but by creating it from nothing.

At the risk of goomba fallacy

1. Guys so AI is a bubble but that doesn’t mean it’s not useful, just like dot com bubble

2. “Someone would have to see value in that. Right now the value prop of AI is that maybe someday it'll get better and we'll be able to build useful, profitable consumer products on it”

Calling OpenAI and Anthropic a pyramid scheme is absurd

In other contexts people would ask me not to cherry pick extreme statements like “OpenAI and Anthropic will go away”. So I don’t have to cherry pick. I just have to pick what you said.

1. GPT-4.5 launch. "AI crash is gonna lead to decade long winter." https://news.ycombinator.com/item?id=43206226

2. Claude Code limits tightened. A commenter expects the bubble to pop, while arguing AI will succeed long term. https://news.ycombinator.com/item?id=44599297

3. GPT-5 disappoints. "Teetering on the edge of a huge bubble pop." https://news.ycombinator.com/item?id=44966583

4. DeepSeek releases a cheaper model. A commenter expects excess data-centre capacity when the bubble bursts. https://news.ycombinator.com/item?id=45413548

5. AMD signs an OpenAI chip deal. A commenter predicts a bubble pop and a flood of cheap GPUs. https://news.ycombinator.com/item?id=45491514

6. Discussion of AI-assisted open-source contributions. A commenter says models must improve soon or the bubble will pop. https://news.ycombinator.com/item?id=45745326

7. Discussion of Opus 4.5 coding. A commenter advises using cheap AI tools before the bubble bursts. https://news.ycombinator.com/item?id=46521192

8. GitHub Copilot moves to usage-based billing. "The bubble has burst." https://news.ycombinator.com/item?id=47929404

9. DeepSeek runs on one AMD accelerator. A commenter predicts a bubble pop will flood the hardware market. https://news.ycombinator.com/item?id=49166790

10. Same DeepSeek thread. "AI is overvalued, so it WILL pop." https://news.ycombinator.com/item?id=49167787

Edit: more examples

1. AI on an old CPU: coding gains don’t justify investment; bubble will burst. https://news.ycombinator.com/item?id=48938412

2. Nvidia–OpenAI deal stalls: GPU bubble, inflated demand. https://news.ycombinator.com/item?id=46834749

3. Lisp machines retrospective: AI bankruptcies and another bubble. https://news.ycombinator.com/item?id=45990309

4. Nvidia customer concentration: AI bubble will pop despite useful technology. https://news.ycombinator.com/item?id=41411258

5. DeepSeek open-source announcement: inflated AI valuations will collapse. https://news.ycombinator.com/item?id=43126535 https://news.ycombinator.com/item?id=43126619

6. OpenAI–AWS deal: spending commitments could bring everything crashing down. https://news.ycombinator.com/item?id=45802550

7. Anthropic funding: final big push before the bubble bursts. https://news.ycombinator.com/item?id=45105333

8. Enterprise AI adoption: disappointing returns will pop the bubble within two years. https://news.ycombinator.com/item?id=48025025

9. OpenAI cash burn: railroad-style bubble and massive investment downturn. https://news.ycombinator.com/item?id=46439964

According to your argument, OpenAI should simply stop R and D which can result in massive jumps in performance because China may catch up?

What is your prescription if I may ask? Don’t invest in research? Let China do it?