The end of the world as we know it!

The end of the world as we know it! AI is here… and it’s lying to you

Most people have probably played around with GPT4 by now. If you haven’t seen it and don’t know what it is, no worries. GPT is supposed to be the next big thing in AI. According to some people, it’s here to take away your job, even if you’re in a field that’s normally thought to be immune from automation. In practice, experts in various fields have said that it’s pretty bad at everything it tries. It’s main real use today is to help programmers with code. The linguistic interface, Chat GPT, leaves a lot to be desired.

ChatGPT isn’t a bad writer. It’s a great simulation of a mediocre writer. It does a pretty decent job of imitating the style of a high school sophomore’s essay writing. That’s not so bad, considering that Americans are reading at a 9th grade reading level today. The big issue is with questions of fact.

Quite frankly, ChatGPT is a liar. The experts call it “hallucination”, in which the AI presents invention or error as fact. For that sophomore writing their essay about Grapes of Wrath (that’s what we had to pretend to read in the 80s, anyway), it’s a problem. And for businesses looking for low-cost solutions for jobs such as phone receptionists and HR interfaces, it’s just unacceptable. Market solutions such as Watson and co. are much better for real world applications.

Beyond that, it’s only effective in English, which is a serious limitation. It doesn’t pretend to be a translator, though. There are already decent tools for that (as long as there is a human at the helm).

Still, if you’re looking for help with something that a real person is going to produce, it has its uses. Without a doubt, it’ll improve with time. It isn’t going to be a Black Swan when some AI tool manages to replicate or improve on human performance in intellectual jobs. We can make simulations in Excel that exceed human decision-making in a lot of cases! But for now, in most applications, GPT is just a curiosity.