Hacker News Reader

Comment by sarchertech | Hacker News Reader

This was an unnecessary experiment that was only done a live mouse (vs a simulation) for the attention grabbing effect.

Obviously an AI agent can perform a simple surgical procedure. The question is how to get it to perform reliably.

The reason AI agents work as well as they do with code is because they get feedback and retry in a loop. Surgery is inherently inflicting “harm” on a patient, so there’s no bright line guardrail to apply to the agent loop.

Agents performing surgery won’t be possible until models get much much more reliable or we develop a simulation that matches reality close enough that we can use that in realtime for their feedback loop. I think we’re likely very far off from either solution.

View on HN

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.

The problems with using an agent for surgery aren’t mechanical. There’s no gain by using a live mouse at this point other than getting attention for the researcher.

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

Let’s agree there’s a 0.000001% chance this study leads someone down a dead end and ends up causing harm.

So I don’t agree that it had any measurable chance at all of helping anyone other than the author.

And if you’re being purely utilitarian, you need to weigh the opportunity cost of what else the money spent on this could have been used for. What else the researchers and peer reviewers could have done with their time, and what they else the mice could have been used for.

>eating meet

I agree this is probably not as bad as you punching babies though.

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

I have interacted with you before and I’d like you to know that you are very unpleasant to talk and you have atrocious debate tactics.

The way you act here makes the world just a little bit worse.

I get that telling you this isn’t likely to be constructive, but if there’s only a 0.000001% chance you reconsider acting in such an unpleasant manner in your next interaction, it was worth it.

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 think there are 2 problems that you have that you should consider addressing.

1. You clearly aren't reading thoroughly before replying. Please go back and reread what I said, and find where you think I said anything about someone getting killed.

2. You are arguing with your perception of the collective consciousness of Hacker News, not discussing something with an individual.

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.

I just happen to think this experiment was stupid and unlikely to provide useful data, and the only reason the researcher used a live mouse was for publicity.

>That’s a non sequitur

The it might somehow kill someone is a non sequitur because you just made that up. My actual argument is that if you say that if you are going to make a utilitarian argument that something provides more benefit than harm overall, you need to weigh the possible benefit against the possible harm.

Your argument was that we can't say for sure there will be a benefit, so lets assume some minute expected benefit.

My argument is that you can't do that without assuming some minute expected harm also exists.

Sure you can essentially ignore the minute expected harm if you believe the expected benefit dwarfs it. But that wasn't your argument, your argument was "let's agree on this minute expected benefit". I believe the expected benefit to be so small as to be immeasurable. And my belief is that it is more likely that the publicity from this result leads other people to perform similar experiments also without any measurable benefit, than that anyone benefits from it.

> 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.

You are completely right, and I'm glad you challenged me. Your aorta is essential, and I should never have suggested otherwise. I sincerely apologize for removing it. Thank you for advocating for yourself.

"Help! I think you injected too much of the drug!"

SAMSA Helpline -If you're addicted to drugs and feel like you need help...

Reply on news.ycombinator.com