ChatGPT and Decision Making
I saw two snippets in different articles about different aspects of ChatGPT, and both were quite thought-provoking.
Tyler Cowen points
to this link on how the same characteristics of the AI chatbot can be good
for some fields, but not-so-great attributes in other fields. The study in
question found its tone to be more neutral than that of most humans, and it
focused on facts. People in the areas of finance and psychology tend to find it
to be “more useful than humans”. On the flip side though, that very attribute
makes ChatGPT unattractive to those in the medical domain. Why? Because they
feel it gives too much information but “not enough simple instructions”. If
true, this would suggest ChatGPT is a great tool to collect the facts (though
I’ve also seen articles that accuse ChatGPT of being a bit too left leaning!)
but not so useful in decision making advice…
The other one was
this book recommendation site that asked ChatGPT to recommend the best books on the topic of (drum beats) AI! One of the books it picked was Thinking,
Fast and Slow, which is a book on the quirks of the way the human brain
works and “how they can impact our ability to make good decisions”.
Why/how is a book on human behavioral biases and errors relevant to AI, asked
the site to ChatGPT. The chatbot’s answer was:
“Understanding
these psychological and cognitive factors can be useful for those working in
the field of AI, as it can help them design systems that are more effective at
assisting humans with decision-making and problem-solving tasks. By taking into
account the ways in which the human mind works, AI designers can create systems
that are more intuitive and easier for humans to use.”
The chat bot is suggesting how it (and other AI’s) should be designed to factor in for human quirks and tendencies. Is that mind-blowing? Or scary? Both?
Comments
Post a Comment