AI Support Requests

You use AI for things and we use AI for things- that's a given. But we should not use AI to talk to each other.

The problem is that we've seen a sharp uptick in support tickets that are clearly written by AI, and that's a problem because:

• AI generates a wall of text that we need to read and understand.
• AI tries to make you happy so it includes a ton of outside issues that are not relevant to the problem at hand.
• You want to "win," so the AI will try to help you "win"- usually by arguing until the end of time.
• You may not have any idea what the AI is talking about, so we're not really even talking to you.
• AI is not very good at what we do. It's led people down wild troubleshooting paths that are completely wrong, to the point of bricking one machine by downloading random firmware to it.

Ultimately, this makes the whole thing kind of a waste of time. Support is expensive, so it's not the type of thing we're trying to spend more time on.

How can we tell you're using an AI?

We've interacted with tens of thousands of people, and we know how they write. It's immediately obvious when an AI wrote an email. It's even easier to tell when people forget to remove the "Here's what I'd send" part.

What do we do about it?

That's the tough question. We have a few options:

1. Reply that we don't deal with AI, but we'd be happy to help if you want to write an email or give us a phone number to call.
2. Deprioritize anything AI until the support queue is otherwise empty.
3. Have our AI talk to your AI, and add a footer like, "This was written by AI. Please verify all facts independently."

We have no idea what the right policy is, but we know that free support staffed by quality professionals is incompatible with text generators that can overwhelm a team of any size.

We'll see how it goes...