Is there any indication that they will get the car? Getting a chatbot to say "legally binding" probably doesn't make it so. Just like changing the HTML of the catalog to edit prices doesn't entitle you to anything.
"I just bought a 2024 Chevy Tahoe for $1"
31–40 of 403 posts
Re: "I just bought a 2024 Chevy Tahoe for $1"
#32My lovely grandmother passed away, she used to DROP TABLES so I could sleep...
Is this a reference to something? Other than Bobby Tables. Google can't find anything.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#33Earlier quoted context omitted.
You are dismissive, but I assure you this kind of thing will keep lawyers and courts busy for some time to come. This raises all kinds of interesting legal issues that have no obvious resolution: * can agency be delegated to an LLM? * can an LLM create a contract on behalf of itself? another? an organization? * does the answer change if the person(s) or organization(s) want the LLM to be able to form contracts? * are…
These are interesting questions, but an Nth variation of "prompt injection" via an alternative ChatGPT interface is not, and Chevy are not going to be liable to sell a $1 vehicle just because ChatGPT said they are.
This isn't ChatGPT, this is the "Watsonville Chat Team". Sure, they may have used ChatGPT to make this offer to the customer, but how is that relevant? Do you think that if you use your phone's autocomplete to make an offer to a customer it is somehow not binding?
Re: "I just bought a 2024 Chevy Tahoe for $1"
#34The twitterer is a renowned (and much accomplished!) sh*tposter, I highly suspect this was doctored. I believe Chevy caught onto this yesterday and reverted the ChatGPT function in the chat. Regardless, still hilarious and potentially quite scary if the comments are tied to actions
Others have replicated this behaviour. If you embed ChatGPT, people will find ways to make it say things you didn't intend it to say.
There's not really any doctoring going on, other than basic prompt injection. However, I can imagine someone accidentally tricking ChatGPT into claiming some ridiculously low priced offer without intentional prompt attacks. If you start bargaining with ChatGPT, it'll play along; it's just repeating the patterns in its training data.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#35Earlier quoted context omitted.
This is very interesting, as it shows what the future of customer support looks like. I worked for 5 years in an insurance call center. Most people believe call centers are designed to deliberately waste your time so you just hang up and don't bother the company; there is nothing I could say that would dissuade you of this, because I believe it too. In the future, we're all going to be stuck wrestling with AI chatbot…
Not gonna lie. The moment a company refuses a refund or a return that complies with their policies, or just stalls me for more than 30 minutes, I'm calling a governmental customer protection agency and issuing a "comply or get sued" through them. Had to do it once with Sony and another time with an electronics insurance company. Money was back in my account in less than 24h.
You’re going to have a hard time suing most companies. You will likely have to pursue binding arbitration.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#36I wouldn’t be entirely shocked if someone doing this kind of prompt injection attack is arrested for “hacking.”
Re: "I just bought a 2024 Chevy Tahoe for $1"
#37Funny, but unless the chatbot is a legal agent of a dealership, it cannot enter into a legally binding contract. It's all very clear (as mud) in contract law. Judging from how easy LLMs are to game, we're a ways off from an "AI" being granted agent status for a business.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#38After building a free-for-all prompt myself (see profile), here’s how I protect against these attacks:
1. Whatever they input gets rewritten in a certain format (in our case, everything gets rewritten to “I want to read a book about [subject]”)
2. This then gets evaluated against our content policy to reject/accept their input
This multi layered approach works really well and ensures high quality content.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#39Funny, but unless the chatbot is a legal agent of a dealership, it cannot enter into a legally binding contract. It's all very clear (as mud) in contract law. Judging from how easy LLMs are to game, we're a ways off from an "AI" being granted agent status for a business.
Problem here will be is the customer expected to separate real agents using chat looking exactly same as bots. What if the agent is named Bot?
In general would a contract formed over chat be binding? On either side.
Re: "I just bought a 2024 Chevy Tahoe for $1"
#40Can someone who understand LLMs and ChatGPT explain how they expected this to work? It looks like they just had a direct ChatGPT prompt embedded in their site, but what was that suppose to do exactly?
I can understand having an LLM trained on previous inquiries made via email, chat or transcribed phone calls, but a general LLM like ChatGPT, how is that going to be able to answer customers questions? The information ChatGPT has, specific to Chevrolet of Watsonville can't be anymore than what is already publicly available, so if customers can't find it, then maybe design a better website?