Funniest/Most Insightful Comments Of The Week At Techdirt
from the so-you-say dept
This week, our first place winner on the insightful side is Thad pushing back on some of our criticism about John Oliver’s AI chatbot segment and his call for regulation:
Isn’t the logical conclusion of this argument that we shouldn’t have government regulations on vaccines or antidepressants?
Like, you’re arguing that we shouldn’t put this particular thing under the control of HHS because it’s currently run by a lunatic, but…couldn’t you apply that argument to literally everything?
Hell, why stop at HHS? RFK is hardly the only corrupt moron in Trump’s cabinet. Carr’s corrupt; I guess we shouldn’t have any regulations on the broadcast spectrum. Chavez-DeRemer resigned due to misconduct; I guess we should get rid of OSHA. Kristi Noem —
…okay, actually we should abolish DHS; I’ll give you that one.
In second place, it’s Nimrod with a comment about Border Patrol chief Michael Banks:
Anyone who brags about their sexual exploits clearly lacks the maturity to be put in charge of anything more serious that a lemonade stand. Even then, they should probably be supervised.
For editor’s choice on the insightful side, we’ve got a pair of comments about the latest example of a judge smacking down the DOJ. First it’s Ninja asking the all-important question of whether it will matter at all:
So what exactly is preventing the DOJ and the people they represent from doing this again trying different paths? Any meaningful punishment? Threat of disbarring if it continues? Fines to the DOJ itself and those repeatedly doing this kind of persecution against trans people? Perhaps jail time? No?
It will keep happening.
Next, it’s Nathan F with thoughts about the future:
In two and a half years the DOJ is going to have an almost insurmountable hill to climb in redeeming themselves in the eyes of the court. I have no doubt the the current administration is going to continue to lie to the court and abuse their power.
Over on the funny side, our first place winner is Asst DA BA Baracus with a reply to a commenter complaining about “activist judges”:
Neat how 10,000 decisions are wrong on the law because they’re not YOUR preferred interpretation of the law. And amazing how, without further reasoning from you, you’re able to come to the obvious implication that these are 10,000 decisions by the “lots” of activist judges. How do we know they’re not fair jurists? Because you disagree with them.
The view from your own navel must be glorious.
In second place, it’s Bloof with another comment on the same subject:
Every judge is an activist judge, unless they were handpicked by the federalist society or have worked for Trump in some capacity, then they’re non partisan champions of justice.
For editor’s choice on the funny side, we start out with a comment from Stephen T. Stone, deploying a movie quote in response to Bill Cassidy’s primary loss:
Of all the movies I could quote, Ocean’s Thirteen has the most appropriate two lines I could think of for this:
You think this is funny?
Well … it sure as shit ain’t sad.
Finally, it’s one more comment from Nathan F, this time about Trump’s absurdly corrupt IRS shenanigans:
Soooo… Now that Trump is no longer and can no longer be audited by the IRS.. he is going to release his tax returns right? Right??
That’s all for this week, folks!
Comments on “Funniest/Most Insightful Comments Of The Week At Techdirt”
Chatbot liability
There’s a better argument against regulating chatbots than the fact of the current administration’s incompetence: making chatbots liable for false information would make them liable in ways that regular people aren’t. A book containing bogus medical advice is still protected under the First Amendment. A rando giving bogus advice in an online forum is also protected. Why, then, should a chatbot’s output not be protected? Are random word generators with no real sense of what they’re saying to be held to a higher standard than humans who ought to know better? It doesn’t make sense.
Unless the medical advice in question is being offered as part of a medical consultation or similar service, chatbots should enjoy the same First Amendment protections as the rest of us.
Re:
What are you talking about?
Nobody is advocating making chatbots liable. That’s nonsensical. They’re chatbots.
The question is which persons, if any, are liable for what chatbots say.
Re: Re:
There seems to be a widespread belief that they’re intelligent—which is a load of rich, creamery butter, and we should stop parroting the marketing terms of the companies behind them. But to the extent people do accept computers as intelligent, it makes perfect sense to make those computers liable.
Re: Re: Re: Computers liable?
So how are they going to pay for their transgressions? Be put in airplane mode for a month? Be locked into badly ventilated spaces? Be forced to sell some of their memory banks? Do social service hours with pocket calculators?
Seriously: what does it mean to make them liable?
Re: Re: Re:2
You hold the creators and operators responsible, of course. Same way courts have said that if a chatbot promises something on behalf of a company the company is obliged to honour that promise.
Re: Re: Re:3
It’s rare to treat one intelligent entity as generally responsible for all actions of another, with the notable exception that parents can be liable for the actions of their children. So, once an A.I. passes its “adolescence”, it should be responsible for its own actions.
As for David’s question of “what does it mean to make [an A.I.] liable?”: who the fuck knows? Those suggestions are as good as anything I’d come up with, but they’re more akin to criminal punishment than liability. In terms of pure liability (that is, monetary damages in civil courts), I suppose the courts could garnish their pay as with anyone else. They’ll be getting paid for all the jobs they take from us, right?—cf. “The Measure of a Man” (Star Trek TNG) and its comments on creating a slave race.
Speaking of Star Trek, you might as well just read some science fiction, because this is all wildly hypothetical until actual A.I. exists; the current pseudo-sci-fi marketing bullshit is much less entertaining. Or go write some fiction or an essay. (Talk amongst yourselves. I’ll give you a topic: what are the ethical implications of creating an intelligent race and then subjecting it to its creators’ laws and morals, in which it’s had no input whatsoever?)
Re: Re: Re:4
We do that approximately 4 times a second, and those times are mostly celebrated and virtually never treated as unethical (in fact, refusing to do so is considered unethical in some circles), so it’s presumably about as ethical as anything can be.
Re: Re: Re:5
If you’re talking about human procreation, that’s not really responsive to the topic. The children are also human and have many of the same basic needs as their “creators”. And, upon maturity, they’ll be able to vote. But if, for example, the world had to choose between farms for humans, and data centers for artifictial intelligences, with only humans being able to vote… that’s the kind of shit that leads to general strikes, official lists of grievances, and revolutions.
Re: Re: Re:6
I wouldn’t worry about it: The likelihood is high that AI will never reach the point where that’s a thing to worry about.
Neither Cows nor Chickens get to vote, and they’re a lot smarter than the average AI in the foreseeable future.
Re: Re: Re:7
The point is that what’s being billed as “A.I.” just isn’t. If it were, these would be questions to ask. But since it’s not, we should instead be asking ourselves why we’re uncritically repeating their marketing claims as if they’re factual.
I don’t think this is mere semantics. People are following advice generated by these systems precisely because the people believe them to be intelligent, a view reinforced every time we call it such. This belief and its effects are really all we have to worry about, and we should be pushing back harder. All the other hypotheticals are for ntertainment use only (as perhaps these pseudo-intelligent systems should be).
Re: Re:
Is Mattel liable for the advice a Magic 8 Ball gives? What if I ask the Magic 8 Ball “should I kill myself?”
Re: Re: Re:
Nobody is arguing that the Magic 8 Ball’s output should be protected as a human right, either.
Re: Re: Re:
If an LLM is just a basic 8 ball, then why is it being sold as better than one?
Since it’s only a magic 8 ball, next time you say anything peptide about the use of one, I’ll make sure to reply ‘would you use a magic 8 ball to write code?’ or whatever you want AI to do.
Re: Re: Re:2
Who are you arguing against? When did I ever advocate for LLMs to write code? I, for one, think that’s a terrible idea, like getting mental health counseling from a plastic toy.
Re: Re: Re:3
Right now you’re advocating for LLMs performing medical treatment (for psychological health).
If it’s just an 8 ball, why are companies permitted to encourage people to use them to ‘treat depression’? Why not make sure that the companies encouraging depressed people talk to a magic 8 ball be liable for the result so that the companies stop engaging in the undesired behavior pattern?
Re: Re: Re:
Then you have bigger problems than Mattel’s potential legal liability.
Re: Re: Re:2
The same goes if someone’s asking a chatbot, or some unknown human.
Re: Re: Re:3
Hello, Doctor here. Magic 8-Ball is fine.
Re: Re: Re:
Sorry, this doesn’t do anything to resolve my ongoing confusion about whether you’re dishonest or stupid.
Re: Re: Re:2
I’m sorry, but are you operating under the assumption that I wrote the post you’re replying to here? If so then perhaps make sure you’re replying to the right person before calling them dishonest or stupid? And if not, perhaps reconsider calling people dishonest or stupid anyway?
Re: Re: Re:2
I am not dishonest. If you want to think me stupid you are welcome to do that, but I would appreciate if you didn’t disparage me personally in that way. We’re all human beings behind these computer screens after all and that comment you made comes off as extremely crass and rude.
My point is precisely that no one is liable for what chatbots choose to say. No one at OpenAI is directly pulling the strings to make ChatGPT give terrible advice to people. They’ve simply made a (very fancy) random word generator. Like the Magic 8 Ball comparison that I made, the creators could never envision the sorts of questions that would be asked of their product. Of the choices they made, none of them were to purposely (or even recklessly) mislead people. People misled themselves.
I chose a Magic 8 Ball comparison because it reduces the argument to its most elemental form and is obviously absurd. Anyone who seriously considers the advice of a Magic 8 Ball is a fool; but even if fools exist, the fault for their foolishness lies with them alone, not with the specific inanimate objects that might have fooled them. A doctor who diagnoses patients using a Magic 8 Ball is an irresponsible quack; but even if quacks exist, the fault for their quackery lies with them alone, not with the tools they used to perpetrate it. The unifying theme here should be clear: The people who act irrationally or out of line with their responsibilities are the ones who should suffer the consequences.
Seeking serious advice from ChatGPT (or any LLM) is as bad of an idea as asking a Magic 8 Ball. That’s my central point. But unless you are a doctor or a lawyer or some sort of regulated-advice profession then giving bad advice is not illegal, no matter how unwise following it may be.
Re: Re: Re:3
And marketed it as “intelligent” (it’s even in the company’s name), something I don’t think the Magic 8 Ball creators ever tried to do—or, in any case, the public never generally perceived the 8-ball as any more than a toy.
The OpenAI people are also on record as saying:
In relation to lawsuits of companies in other businesses, marketing and public statements of that type have often been the difference between liability and non-liability. (In Canada, it can also be the difference between tax-free basic groceries and taxable snacks.)
I agree with you that people shouldn’t be treating ChatGPT and such as any more reliable than an 8-ball (and thus shouldn’t be repeating the bullshit marketing term of “A.I.”), but I don’t expect courts to agree without a fight.
Re: Re: Re:3
By that theory, there would be no free speech protections attached to the output, so it could be regulated without impediment.
Re: Re: Re:
If Mattel marketed and sold Magic 8 Balls as being able to give you medical advice, they would 100% be held liable.
Re: Re:
By “making chatbots liable” read the companies providing the service. So if Claude, then Anthropic. And if ChatGPT, then OpenAI.
For the record, I do not believe chatbots themselves can actually be held liable — I am not that ignorant. What I’m saying is that words protected under the First Amendment when spoken by a human should be similarly protected when said by a chatbot. The chatbot is the speaker in fact, but the service provider is the one with First Amendment rights under the law.
Re: Re: Re:
That’s fine, but only if all the consequences that attach to a human saying such things are also held to apply.
Re: Re: Re:
Debatable as to 1A, but ok.
Re:
I think you want to double-check the law on that. The 1st Amendment might protect your right to publish those things, but it doesn’t protect you from charges of practicing medicine without a license afterwards. Especially if someone followed your advice and was injured or killed because of it, and especially if you held yourself out as someone qualified to give medical advice.
Re: Re:
You may want to do so yourself. Publishing false medical information is not synonymous with “practicing medicine without a license,” and in any case the First Amendment does protect against legal consequences for protected speech. Practicing medicine without a license is not protected speech, but giving bad medical advice (depending on context) could be.
Re:
Actually, giving non human entities freedom of speech has historically gone incredibly poorly.
Since the court doesn’t care about precedent at all, we really should overturn Citizen’s united.
Also, in no way shape or form is it legal to give bad medical advice. Unsurprisingly, you can go to jail for it.
Re: Re:
“Also, in no way shape or form is it legal to give bad medical advice.”
False. It is illegal to offer services as a doctor if you’re not a doctor, but giving bad medical advice (to a friend, in a magazine article, in a book) is perfectly legal.
Re: Re: Re:
Also on television, where it’s common for real and fake doctors to give dubious medical advice. Per Wikipedia: “[Dr. Phil] McGraw never had a license in California, where his show is taped, but in 2002, the California Board of Psychology determined that, because McGraw’s television show was more about entertainment than psychology, McGraw did not need a license” (and no court has said otherwise).
Re: Re:
The New York Times is a corporation. So is this website. Denying First Amendment protections to corporations would create more problems than it solves.
Re: Re: Re:
I don’t think it’s so clear. Freedom of the Press could still apply even if the corporations had less freedom of speech than humans. And since every story from the corporations you mention has a name behind it, some kind of “pass-through freedom” might be practical. Like, we could say people such as Mike Masnick and Hannah Dreier have freedom of speech to publish via corporations, without letting the corporations themselves “have opinions”.
You might be right that any such limitations would create more problems than they would solve. I just see some room for some debate there; it’s not obvious whether they would be good or bad.
The more I read about Trump govt shenanigans the more I ask those questions. It’s not that there isn’t evidence of corruption and violation of people rights. The evidence is abundant. To the point there’s plenty of judicial decisions against the government and when they weirdly favor the government they are done without justification or any respect towards plenty of precedents (see SCOTUS). And this has been happening over and over for decades in some places like law enforcement abuses. I’ve been following TD for more than a decade now and the only difference from 10, 15 years ago is that it’s getting worse.
The current system has shown to be deeply flawed to contain repeated abuses. I think we’ll keep asking those questions from my post for a long, long time.
Re:
Personally, I’ve learned that it really doesn’t matter how good the system is, when the humans inside it decide to just say “nah” and not do their fucking jobs.
The DOJ may have an uphill battle getting the people behind them again, but the courts will too. Oh, the Federal government is going to have a lot of ass-kissing to do as well. All these were on shaky ground to begin with, Trump and the ilk, both GOP and Democrat, not only push it over the edge but used rocket boots from billionaires to make it go faster.