I've had decent luck with saying straight out at the start of the call "I know this is a scan. I am on the federal Do Not Call list and will be reporting you. I'm hanging up now" and then hanging up. My intention is to get my number marked as having a low probability of converting; if you lead them along they might think a few more calls will give them a decent chance at profiting. I also feel sympathy for people whose best option is telephone scams.

Note: I'm lying about Do Not Call. It's just another way for people to learn your number has a real person behind it. But the synchronization between the Federal list and internal systems of callers is a shitshow so actually reputable mass dialers will just assume they missed you and add you to their internal list. (Edit: So of course I don't report people. Even if I was really on the list, that takes time and has essentially no return)

Doesn't this signal to the malicious caller that your number is an active line with a human willing to pick up the call? I imagine that information is valuable to malicious callers who probably sell lists of active phone numbers of curious humans willing to pick up.

I heard that advice 10-20 years ago, and I think it made sense then: junk calls were accomplished with autodialers feeding into phone banks with real humans to finish the scam.

With modern VOIP I’d be surprised if anyone compiles lists of possible marks: your outbound traffic can be way more than before, so why not let the computer call everyone and see who bites every time?

Is there some way to implement this as a service with a chat bot or just prerecorded mumbling and confused statements? I guess it could happen by just transferring the call.

The true future of AI is robots sending eachother spam.

That would be Lenny: https://shaun.net/notes/introducing-lenny/