The “AI” Challenge – My Response
Kevin has been kind enough to respond to my challenge on why “artificial intelligence.” He notes that the current chatbots have some uses but mostly is looking forward to “artificial general intelligence,” which is some time away. It will be able to do many good things if it turns out the way he expects it to.
Kevin also lives with two adorable cats. In the picture of my cats that he used, it is, left to right, Zooey and Ric.
Both our posts got lots of comments, many volunteering yet more things that “artificial intelligence” might do. As I read the comments and Kevin’s response, though, I realized that what I really wanted to know was what the developers of these models think they are doing. None responded, so I still don’t know, although there were many good guesses in the comments.
Machine translation, a more limited machine learning app, has come a long way, but it still has some way to go. The accuracy of translation into English decreases as the language becomes less like English. So it works well for French, less so for German, and less so for Russian. Estonian is a Finno-Ugric agglutinative language, and translations have improved quite a bit but can be pretty bad, too. It would be interesting to know if translation between Estonian and Finnish is as good as that between French and English. Syntax is part of this and supposedly something that the machine language folk are particularly proud of.
Even without “AI,” I’ve been very pleased with Target’s customer service over the past three years. I’ve continued to shop there without ever setting foot in the store. I order on the internet and am notified when the order is ready or if items will be shipped. I tell them when I’m coming to pick up the order, and they have it ready. Machine dialog refunded one item that wasn’t in an order and sent a replacement for one that got lost in shipping. None of this was difficult, for me or for the system.
Now ChatGPT4 has been rolled out to do its damage in the world. I worry primarily about two things: 1) that too many humans will believe that the chatbot is “thinking” or “sentient” and 2) that garbage from chatbots we know fabricate stuff will overrun the internet. One of our commenters recommended this article on Emily Bender to me. It is good to know she is in the middle of things, fighting the good fight. I can leave it to her and continue to follow this issue rather cursorily.
As we’ve seen over the weekend, the people who fund these things are neither very bright nor emotionally stable. The company behind ChatGPT4 fired their ethics team. We have real problems to address – global warming and poverty to name a couple – and now we have another inflicted upon us by that bunch.