Why are chatbots (still) so bad?

Getting a human-like response from a software system is a very difficult problem to solve. We're not there yet, not even close.

I think a more relevant question is - why do so many chatbots exist in the wild considering how bad they are?

I was doing some teaching about chatbots recently and one of the key points was passing over to a human when needed.

The problem is one of expectation. Companies want chatbots to do away with humans but users want chatbots to be like humans.

Maybe we should call them faqbots and remove the spoof pretty girl icon :-)

Because what niche they’ve carved out isn’t driven by utility for end users but by cost reduction for the company in reduced headcount.

They add value, just not to anyone who has to actually deal with them.