> Because they should 100% be liable for the latter.
Why? I don't see that a drug designed by ChatGPT should result in any more or less liability than a drug designed by a human?
I think if a human designs a drug and tests it and it all seems fine and the government approves it and then it later turns out to kill loads of people but nobody thought it would... that's just bad luck! You shouldn't face serious liability for that.
If we start from the position of the marketing hype and even Sam Altman's statements, these tools will "solve all of physics". To me it's laughable, but that's also what's driven their outsized valuations. Using the output to drive product decisions and development, it's not hard to imagine a scenario where a resulting product isn't fully vetted because of the constant corporate pressure to "move faster" and the unrealistic hype of "solve all of physics". This is similar to Tesla's situation of selling "Full Self-Driving" but it actually isn't in the way most people would understand that term and so they lost in court on how they market their autonomous driving features.
Can't agree with this. No, not at all. That can't be true... That's not "just bad luck". I believe this is actually a serious case of negligence and oversight - regardless of where exactly it occurred, whether on the part of the drug’s manufacturer, the government agency responsible for oversight, or somewhere else. It just doesn’t work that way. Any drug undergoes very thorough and rigorous testing before widespread use (which is implied by "millions of deaths"). Maybe I’m just dumb. And yeah, this isn’t my field. But damn it, I physically can’t imagine how, with proper, responsible testing, such a dangerous "drug" could successfully pass all stages of testing and inspection. With such a high mortality rate (I'll reinforce - millions of deaths cannot be "unseen edge cases"), it simply shouldn’t be possible with a proper approach to testing. Please, correct me if I’m wrong.
> I don't see that a drug designed by ChatGPT should result in any more or less liability than a drug designed by a human?
It’s simple. In this case, ChatGPT acts as a tool in the drug manufacturing process. And this tool can be faulty by design in some cases.
Suppose, during the production of a hypothetical drug at a factory, a malfunction in one of the production machines (please excuse the somewhat imprecise terminology) - caused by a design flaw (i.e., the manufacturer is to blame for the failure; it’s not a matter of improper operation), and because of this malfunction, the drugs are produced incorrectly and lead to deaths, then at least part of the responsibility must fall on the machine manufacturer. Of course, responsibility also lies with those who used it for production - because they should have thoroughly tested it before releasing something so critically important - but, damn it, responsibility in this case also lies with the manufacturer who made such a serious design error.
The same goes for ChatGPT. It’s clear that the user also bears responsibility, but if this “machine” is by design capable of generating a recipe for a deadly poison disguised as a “medicine” - and the recipe is so convincing that it passes government inspections - then its creators must also bear responsibility.
EDIT: I've just remembered... I'm not sure how relevant this is, but I've just remembered the Therac-25 incidents, where some patients were receiving the overdose of radiation due to software faults. Who was to blame - the users (operators) or the manufacturer (AECL)? I'm unsure though how applicable it is to the hypothetical ChatGPT case, because you physically cannot "program" the guardrails in the same way as you could do in the deterministic program.
> I physically can’t imagine how, with proper, responsible testing, such a dangerous "drug" could successfully pass all stages of testing and inspection.
It might cause minor changes that we don't yet know how to notice, and which only cause symptoms in 20 years' time, for example. You can't test drugs indefinitely, at some point you need to say the test is over and it looks good. What if the downsides occur past the end of the test horizon?
> ChatGPT acts as a tool in the drug manufacturing process. And this tool can be faulty by design in some cases.
ChatGPT is not intended to be a drug manufacturing tool though? If you use any other random piece of software in the course of designing drugs, that doesn't make it the software developer's fault if it has a bug that you didn't notice that results in you making faulty drugs. And that's if it's even a bug! ChatGPT can give bad advice without even having any bugs. That's just how it works.
In the Therac-25 case the machine is designed and marketed as a medical treatment device. If OpenAI were running around claiming "ChatGPT can reliably design drugs, you don't even need to test it, just administer what it comes up with" then sure they should be liable. But that would be an insane thing to claim.
I think where there may be some confusion is if ChatGPT claims that a drug design is safe and effective. Is that a de facto statement from OpenAI that they should be held to? I don't think so. That's just how ChatGPT works. If we can't have a ChatGPT that is able to make statements that don't bind OpenAI, then I don't think we can have ChatGPT at all.
> It might cause minor changes that we don't yet know how to notice, and which only cause symptoms in 20 years' time, for example.
In that case, even if it leads to many deaths, it would be difficult - if not practically impossible - to hold anyone accountable, even if it were possible. However, such a turn of events is difficult, or rather, practically impossible to predict, don’t you think? I apologize for not clarifying this point in my original comment, but I wasn’t referring to delayed effects - I was referring to what becomes evident almost immediately (for example, let’s say “within a year and a half at most”) after the drug is used. Yes… I’m sorry, I just didn’t phrase my thought correctly. I apologize for that.
> ChatGPT is not intended to be a drug manufacturing tool though?
That’s certainly the case right now. However, LLMs like GPT, Claude, Gemini, and others weren’t created for waging war, were they? Then why did Anthropic recently have - let’s just say... "some issues in its relationship" with the DOD, if they were not involved in this, if Claude was not meant to be used in war? Why was the ban on using Gemini to develop weapons removed from its terms of service?
You’re right that LLMs weren’t created for such purposes, and to be honest, I believe that - at least for now - it’s simply unethical to use them for that. These aren’t the kinds of decisions and actions that should be outsourced to a machine that bears no responsibility - moral or legal.
> ChatGPT can give bad advice without even having any bugs. That's just how it works.
To continue my thought, this is precisely why I believe it is unethical to give LLMs any tasks whatsoever that involve human lives. There are simply no safety guarantees - not just "some", but none at all - aside from unreliable safety fine-tuning and prompting tricks. For now, that’s all we can count on.
> If OpenAI were running around claiming "ChatGPT can reliably design drugs, you don't even need to test it, just administer what it comes up with" then sure they should be liable. But that would be an insane thing to claim.
They don't claim it yet. And, as one person (qsera) mentioned below your comment:
> The trick is to make people behave like that without actually claiming it. AI companies seems to have aced it.
They probably won't claim exactly that "ChatGPT can reliably design drugs", just because of the possible consequences. But I'm almost certain there will be something similar in meaning, though legally vague - so that, from a purely legal standpoint, there won't be any grounds for complaint. What's more, they are already making some attempts - albeit relatively small ones so far - in the healthcare sector; for example, "ChatGPT Health"[1]. I don't think they will stop there. That's a business after all.
> if ChatGPT claims that a drug design is safe and effective
I have already said before that the OpenAI will not be the only one who should be held responsible in this case. The (hypothetical) user should also bear some responsibility, and in the scenario you described, the primary responsibility should indeed lie with them. That said, I may be wrong, but it’s possible to fine-tune the model so that it at least warns of the consequences or refuses to claim that "this works 100%". This already exists - models refuse, for example, to provide drug recipes or instructions for assembling something explosive (specifically something explosive, not explosives - I recently asked during testing, out of curiosity, Gemma 4 how to build a hydrogen engine - and the model refused to describe the process because, as it said, hydrogen is highly flammable and the engine itself is explosive), pornography, and things along those lines. Yes, I admit, it’s far from perfect. But at least it works somehow. By the way, if I’m not mistaken, many models even include disclaimers with medical advice, like "it’s best to consult a doctor".
In short, what I’m getting at is that the issue lies in how convincing the LLMs can be at times. If it honestly warns of the dangers of using it, if it says "this doesn’t work" or "this requires thorough testing", and so on, but the user just goes ahead and does it anyway - well, that’s like hitting yourself on the finger with a hammer and then suing the hammer manufacturer. It’s a different story when the model states with complete confidence that "this will definitely work, and there will be no side effects" - and user believes it; there should be some effort put into preventing such cases. But otherwise, yes, I think you’re right about the scenario you described.
And to conclude - I don’t think that when it comes to drug development, we’re talking about ordinary people or individual users. In the context of the parent post, it is implied (though I may have misunderstood) that ChatGPT would be used by entire organizations, such as pharmaceutical companies - just as LLMs in a military context are used not by individuals, but by the DOD and similar organizations. I think this shifts the level of responsibility somewhat. Because when OpenAI enters into a contract for the use of its product, ChatGPT, in the process of drug development and manufacturing, it’s kind of implied that ChatGPT is ready for such use.
EDIT: I'm sorry that my reply is so long, I'm just trying to express all of my thoughts in one which is probably not a good thing to do. I would write something like a blog post about that, but there's a lot written about this topic already, so...
Yeah, and I have also used translator in some parts because English is not my native language.
> it simply shouldn’t be possible with a proper approach to testing.
It just has to be delayed. Like many years after application. Or trigger on very specific and rare circumstances. Not likely in a trial, but near certain at a population scale.
Or both...
On top of that, If I remember correctly, this liability wavering also exist for Vaccines.
> It just has to be delayed. Like many years after application.
That's one thing. In this case, I don't really know if it's possible to test for something like delayed effects. I'm not even sure if you can identify them with 100% certainty; if you can prove that these effects come from this particular drug and not from another one.
> Or trigger on very specific and rare circumstances. Not likely in a trial, but near certain at a population scale.
And this is different thing. "Specific and rare circumstances" will not lead to millions of deaths (I apologize if I’m being too nitpicky about this particular phrasing, but I want to speak specifically in the context of “millions of deaths”). “Specific and rare circumstances” occur even with fully effective and "proper" medications - this is called “contraindications.” But such rare cases, as I’ve already said, will not lead to mass deaths - precisely because they are rare. I apologize again for focusing on the "millions", but please don’t confuse the scale of the problem.
Why? I don't see that a drug designed by ChatGPT should result in any more or less liability than a drug designed by a human?
I think if a human designs a drug and tests it and it all seems fine and the government approves it and then it later turns out to kill loads of people but nobody thought it would... that's just bad luck! You shouldn't face serious liability for that.