"This is a remarkable submission," Civil Resolution Tribunal (CRT) member Christopher Rivers wrote.""
From https://www.cbc.ca/news/canada/british-columbia/air-canada-c...
But, to your question, my guess is that would basically be telling people not to avoid their chatbot, which they don't want to do.
Obviously that wouldn't fly. So why would it fly with the AI chatbot's advertising discounts?
You wouldn't normally expect an AI chatbot to be authorized to make offers. Its purpose is to try to answer common questions and it has been widely covered in popular media that they hallucinate etc.
I think only software engineers would think this. I don't think it is obvious to a layperson who probably has maybe never even used one before.
And they did get sued. Next time maybe they'll make sure software they connect to their website is more reliable.
They didn't program it to do that, it's a characteristic of the technology that it makes mistakes. Which is fine as the public learns not to blindly trust its answers. It seems silly to assume that people won't be able to figure that out. People are capable of learning how new things work.
This is like the people who set the cruise control in their car when it first came out and then climbed into the back of the car to take a nap. That's not how it works and the technology isn't in a state where anybody knows how to do better.
There's absolutely a technology available to make a chatbot that won't tell lies: connect a simple text classifier to a human-curated knowledge base.
The result would be a de facto ban on AI chatbots, because nobody knows how to get them not to make stuff up.
> I'm glad they're not allowed to use such unreliable, experimental technologies in their airplanes (737 Max notwithstanding).
If you use unreliable technology in an airplane, it falls out of the sky and everybody dies. If you use it in a chatbot, the customer can e.g. go to the company's website to apply for the discount it said exists and discover that it isn't there, and then be mildly frustrated in the way that customers commonly are when a company's technology is imperfect. It's not the same thing.
> There's absolutely a technology available to make a chatbot that won't tell lies: connect a simple text classifier to a human-curated knowledge base.
But then it can only answer questions in the knowledge base, and customers might prefer an answer which is right 75% of the time and can be verified either way in five minutes than to have to wait on hold to talk to a human being because the less capable chatbot couldn't answer their question and the more capable one was effectively banned by the government's liability rules.
No, the result would be a de facto ban on using them as a replacement for customer service agents. I support that for the time being since AI chatbots can't actually do that job yet because we don't know how to keep them from lying.
They could put a disclaimer on it of course. To be sufficiently truthful, the disclaimer would need to be front and center and say something like "The chat bot lies sometimes. It is not authorized to make any commitments on behalf of the company no matter what it says. Always double-check anything it tells you."
That's the point of it -- you don't have to wait on hold for a human to get your answer, and you could plausibly both receive it and validate it yourself sooner than you could get through to a human.
But what does that even mean? If Ford trains a chatbot to answer questions about cars purely for entertainment purposes, or to get people excited about cars, a customer could still use it for "customer service" just by asking it questions about their car, which it might very well be able to answer. But it would also be capable of making up warranty terms etc., so you've just banned that thing and anything like it.
> I support that for the time being since AI chatbots can't actually do that job yet because we don't know how to keep them from lying.
It's pretty unlikely we could ever keep them from lying. We can't even get humans to do that. The best you could do is keep them on a script, which is the exact thing that makes people hate existing human customer service reps who can't help them because it isn't in the script.
> To be sufficiently truthful, the disclaimer would need to be front and center and say something like "The chat bot lies sometimes. It is not authorized to make any commitments on behalf of the company no matter what it says. Always double-check anything it tells you."
Which is exactly what's about to start happening, if that actually works. But that's as pointless as cookie banners and "this product is known to the State of California to cause cancer".
I expect something that's presented as customer service not to lie to me about the rebate policy. As long as what it says is plausible, I expect the company to be prepared to cover the cost of any mistakes, especially if the airline only discovers the mistake after I've paid them and taken a flight. Compensating customers for certain types of errors is a normal cost of doing business for airlines, and the $800 CAD this incident cost the airline is not an exorbitant amount. The safety valve here is that judges and juries do test against whether a reasonable person would believe a stated offer or policy; I can't trick a chatbot into offering me a billion dollars for nothing and get a court to hold a company to it.
If Ford presents a chatbot as entertainment and makes it really clear at the start of a session that it doesn't guarantee the factual accuracy of responses, there's no problem. If they present it as informational and don't make a statement like that, or hide it in fine print, then it says something like "the 2024 Mustang Ecoboost has more horsepower than the Chevrolet Corvette and burns less gas than the Toyota Prius", they should be on the hook for false advertising to the customer and unfair competition against Chevrolet and Toyota.
Similarly, if Bing or Google presents a chatbot as an alternative to their search engine for finding information on the internet, and it says "Zak's photography website is full of CSAM", I'm going to sue them for libel.
Sure, but a billion people could each trick it into offering them $100, which would bankrupt the airline.
> they should be on the hook for false advertising to the customer and unfair competition against Chevrolet and Toyota.
But all you're really doing is requiring everyone to put a banner on everything that says "for entertainment purposes only". Because if something like that gets them out of the liability then that's what everybody is going to do. And if it doesn't then you're effectively banning the technology, because "have it not make stuff up" isn't a thing they know how to do.
If that means companies can't use chatbots to replace customer service agents yet, so be it.
But what does that matter? So someone posts on Reddit how to trick the chatbot into offering a rebate and then 75% of their customers have done it by the time they realize what's going on and now they're out of business.
> If that means companies can't use chatbots to replace customer service agents yet, so be it.
You're still not articulating any way to distinguish "customer service" from any other functioning chatbot. A general purpose chatbot will answer customer service questions, so how does this not just ban all of them?
If, instead of a chatbot, this was about incompetent support reps that lied constantly, would you make the same argument? "We can't hire dirt-cheap low-quality labor because as company representatives we have to do what they say we'll do. It's so unfair"
If Microsoft puts ChatGPT on Twitter so people could try it, and everybody knows that it's ChatGPT, and then it started offering companies free Windows licenses, why should they have to honor that? It's obvious why it might do that but the purpose of letting people use it wasn't so it could authorize anything.
If the company holds a conference where it allows third party conference speakers to give talks, which everybody knows are third parties and not company employees, should the guest speakers be able to speak for the company? Why would that accomplish anything other than the elimination of guest speakers?
I think that banning lying to customers is fine.
It sounds like you meant to say that they didn’t _intentionally_ program it to do that. They didn’t find the system under a rock and unleash it on the world; they made it.
Yes. And therefore people should be able to assume that the answers are correct.
Some people have heard of ChatGPT, and some of those have heard that they hallucinate, sure. But that's still not that many people. And they don't know that a question answering chat bot like this is the same technology!
Why is that a necessary requirement? Something can be useful without it being perfect.
If I have to double-triple check elsewhere to make sure that the chatbot is correct, then what's the point of using the chat bot in the first place? If you can't trust it 99% of the time, or if the company says "use this, but nothing it says should be taken as fact", then why would i waste my time?
Awhile back another commenter called them a “demented Clippy” which about sums them up for me.
Because you can ask it a question in natural language and it will give you an answer you can type into a search engine to see if it's real. Before you didn't know the name of the thing you were looking for, now you do.
> If you can't trust it 99% of the time, or if the company says "use this, but nothing it says should be taken as fact", then why would i waste my time?
The rate at which it makes stuff up isn't 99%, is the point. For common questions, better than half of the answers have some basis in reality.
What you're describing isn't what you expect, it's what you wish were the case even though you know it isn't.
Will Air Canada be legal for my friend going against company policy?
What they told it to do was to behave very unpredictably. They shouldn’t have done that.
These ones do what they "learned" from a lot of input data using a process that is us mimicking how we think brains could maybe function (kinda/sort off with a few unbiological "improvements").
Say your webserver isn't scaling to more than 500 concurrent users. When you add more load, connections start dropping.
Is it because someone programmed a max_number_of_concurrent_users variable and a throttleExtraAboveThresholdRequests() function?
No.
Yes, humans built the entire stack of the system. Yes every part of it was "programmed", but no this behaviour wasn't programmed intentionally, it is an emergent property arising from system constraints.
Maybe the database connection pool is maxed out and the connections are saturating. Maybe some database configuration setting is too small or the server has too few file handles - whatever.
Whatever the root cause (even though that cause incidentally was implemented by a human if you trace the causal chain back far enough) this behaviour is an almost incidental unintended side effect of that.
A machine learning system is like that, but more so.
An LLM, say, is "parsing" language in some sense, but ascribing what it is doing to human design is pretty indirect.
In a way you typing words at me has in some way been "programmed" into you by every language interaction mankind has had with you.
I guess you could see it that way, but I don't think it's a particularly useful point of view.
In the same way an LLM has been in directly "programmed" via it's model architecture, training algorithm and training data, but we are nowhere near the understanding of the process to be able to consider this "programming" it yet.
Its behavior incorporates randomness and is unpredictable and hard to keep within bounds on purpose and they decided to tell a computer to follow that unpredictable instruction set and place it in a position of speaking for the company, without a human in between. They shouldn’t have done that if they didn’t want to end up in this sort of position.
This is also a management failure in badly evaluating and managing the risks of a new technology.
We disagree in that I don't think that its behaviour being hard to predict is on purpose: we have a new technology that shows great promise as tool to work with language input and outputs. People are trying to use LLMs as general purpose language processing machines - in this case as chat agents.
I'm reacting to your comment specifically because I think you are evaluating LLMs using a mental model derived from normal software failures and LLMs or ML models in general are different enough to make that model ineffective.
I almost fully agree with your last comment, but the
> they decided to tell a computer to follow that unpredictable instruction set
reflects what I think is now an unfruitful model.
Before deploying a model like this you need safeguards in place to contain the unpredictability. Steps like the following would have been options:
* Fine-tuning the model to be more robust over their expected input domain,
* Using some RAG scheme to ground the outputs over some set of ground truths,
* Using more models to evaluate the output for deviations,
* Business processes to deal with evaluations and exceptions, Etc
The legal system recognizes that people, or groups of people, are subject to legal authority. This is a story about a piece of software Air Canada implemented which resulted in them posting erroneous information on their website.
It is very different than if an employee were to, in writing, make a statement that a reasonable person would find reasonable.
If the chatbot told them that they'd get a billion dollars, the courts would not hold Air Canada responsible for it, just as if a programmer put a decimal wrong and prices became obviously wrong. In this case, the chat bot gave a policy within reason and the court awarded the passenger what the bot had promised, which is a completely correct judgement.
If they decide it is reliable enough to be put in front of the customer, they must accept all the consequences: the benefits like having to hire less, and the cons, which is that they have to make it work correctly.
Otherwise, woopsy, we made our AI handle our accounting and it cheated, sorry IRS. That won't fly.
There is a common misconception about law that software engineers have. Code is not law. Law is not code. Just because something that looks like a function exists, you can't just plug in any inputs and expect it to have a consistent outcome.
The difference between these two cases is that even if a chat bot promised that, the judge would throw it out, because it's not reasonable. Also, the firm would have a great case against at least the CS rep for this collusion.
If your friend of a CS agent promised you a bereavement refund (As the chatbot did), even though it went against company policy, you'd have good odds of winning that case. Because the judge would find it reasonable of you to believe and expect that after speaking to a CS rep, that such a policy would actually get honored. (And the worst that would happen to the CS rep would be termination.)
If your friend promised you something reasonable in the course of carrying out their duties, and you honestly believed them, I think that would be legal and enforceable just as this case suggests.
If a random AC employee gave you a free flight, on the other hand, you'd be entitled to it.
Anyway, the chat bot has no agency except that given to it by AC; unlike a human employee, therefore, its actions are 100% AC actions.
I don't see how this is controversial? Why do people think that laws no longer apply when fancy high-tech pixie dust is sprinkled?
The company would be entirely within their rights to say 'this employee was wrong, that is not our policy, goodbye!'. This happens all the time with more minor incidents.
This chatbot merely said something was possible, no legally binding agreement occured.
So there was no contract but a consumed flight. The court has to retroactively figure out a reasonable contract in such cases. That Air Canada couldn't just apply the reduced rate once they learned of their wrong communication marks them as incredibly petty.
If you were standing at the customer service desk, and instead they said: "sorry about the delay, your next two flights are free", then all of a sudden this is "reasonable".
[0] perhaps because they're disgruntled and trying to hurt their employer
[1] generative models are not an exception; they're a way of telling computers to generate text that sometimes contains falsehoods
I'm sure that if the bot had said that the airline would raise your dead relative from the grave and make you king of the sky or something equally unbelievable the courts wouldn't have insisted Air Canada cast a crown and learn necromancy.
And if it was a billion dollars?
The chatbot instructed the passenger to pay full price for a ticket but stated they could get a refund later. That refund policy was a hallucination. The victim her just walked away with a discounted ticket as promised not a billion dollars.
So replacing all their customer support staff with AI that misleads customers is OK? That's pants on head insane, so why spend time trying to justify it.
One major difference is the AI wasn't your friend, another is that you didn't get it hired at Air Canada, another is that the promise wasn't $1B, etc...
If you are promised something reasonable by an agent of the company who you are not conspiring with, then the company is bound to follow through on the promise because you do have a reasonable expectation that what they are telling you is the real policy.
("Chatbot says you can submit a form within 90 days to get a retroactive bereavement discount" sounds perfectly reasonable, so the doctrine applies.)