Embarrassing? It's an accomplishment for them! They're saving the company a huge amount of money. Employees are expensive, even the ones in the third-world you can pay $2/hr.
Exactly, these diffuse international entities are merely responding to structures of incentives which are ultimately the government’s responsibility to create and enforce.
Contrary to the quote in the article, proper investment in response and mitigation has, so far, been proven to ~not~ be mandatory for making incredible amounts of cash, which is the primary objective of these behemoths.
Now that externalized costs are starting to personally affect members of government and enforcement, I hope we’ll see some meaningful regulation around accounts on private-but-essential digital platforms resembling a general bill of rights rather than letting private companies get away with nearly anything by default, as is the current state. However merely asking these pseudo-nation-states masquerading as companies to behave differently and expecting to be taken seriously is good for a laugh!
Oddly I thought this niche would have been the first target for LLM's.
It could seriously outperform human support for some queries.
The user claims their account was stolen, if they fill the report and attempted to log in from a known location/ip using a previously valid password you can immediately block unusual spending and use the attempt to further confirm they lost control over their account. Can quickly gather all kinds of fuzzy data, do they still control their phone/pc? Are they sending dubious messages?
The real fun begins where scammers try to do similar things simultaneously with multiple stolen accounts. If you bother to gather the data the patterns are much easier to find.
Given that Canada Airlines killed their Chatbot because they were legally held accountable for the policy their chatbot had created in their case, I think many of these companies are thinking twice before allowing an LLM to respond to people out of fear of the words of the chatbot making legal issues for them.
Killing it may have been more of a PR move than anything. Airlines have regularly been caught in incorrect prices listed on their systems. This is not new. Searching for them is basically a sport. They do just fine with a few incorrect prices flowing through the system.
Aliexpress has delivery companies, vendors and a complaint system that are total chaos. Which ends in disputes making it into credit card and paypal disputes systems. That probably still works just fine for them.
Companies regularly claim that "oops, that employee made a mistake in telling you that" and I suspect most customers let them.
So that I expect it's going to be the result: companies will use chatbots, will try to strong arm customers into "correcting chatbot mistakes" and in the end live with a few results against them.
It can serve canned responses np. If there is a need to freestyle it has to properly explain it is a chat bot and that the response isn't legally binding.
You can pretty much say anything if you put a question mark behind it?
