The limitations of AI
Much handwringing is going on these days about the impact of Artificial Intelligence on humanities’ future. What jobs will be lost? Will AI rise up to enslave the human race. There are many concerns about AI’s impact on human society.
However, I recently read an article by Dan Foster titled, Artificial Intelligence is Coming For Religion. The writer discusses the impact of AI on religion, in particular Christianity. In this article, he tells us that he tested the response of ChatGPT (an AI writing device) to the question,”If God Is So Good, Why Do Bad Things Happen?”
In addition to several other explanations, AI answered discussing the concept of free will. In essence AI gave the traditional answer that evil happened because humans had free will and chose to do evil things. It wasn’t God’s action that caused evil, but human choice. It also went on to discuss other possibilities for evil occurrences.
This answer about human choice showed me that maybe AI was not the boogeyman we make it out to be. It showed me something that gave me hope. Could AI be limited and inadequate in its ability to answer and solve problems?
I found the this AI response to the question of evil inadequate. The answer of free will causing evil is traditional. Nothing new here. It doesn’t discuss the deeper questions of free will and the VICTIMS of evil, in other words, where was their (the victims) freedom? Where was the freedom of the 6 million victims of the evil perpetrators of the holocaust? They surely did not choose to climb on boxcars and ride to concentration camps.
Ai only gives the traditional answer to this question. It did not show the ability to extend its thinking into other possibilities such as the freedom of victimhood. It seems at least on the surface to be limited in its answers. Ai seems lacking in its attempt to address more than superficial answers.
Maybe at some point, AI will evolve into a true monster that can extend its thinking. But for now, it seems an impotent monster, although we should not blind ourselves into thinking this is just another toy.
While I have not tried ChatGPT, I have read a bit of the responses that Ian Hobson published from his weeklong chat with the software and I have found that it basically searches the Internet for the answers requested and puts together plausible sentences based on the prominence of various answers. Certainly, buzzwords or hot topics would get higher ranking when it comes to answering the questions. As a writer, I'm sick and tired of the word, "impact," being misinterpreted and misused. impacted by early education intervention.
I find that the software used by companies to avoid answering the phone are totally inadequate except for the simplest of requests like, What's my balance? Or, What do I owe? For people too lazy to look it up on the appropriate website. While trying to get a serious question or request answered, I find the use of these systems waste a lot of our time trying to get to the right person to solve the problem or answer correctly what's needed.
ChatGPT has been singled out as left-wing by some critics, probably because of the straightforward answers you get when asking about religious dogma and beliefs. I don't see this as getting anywhere near the capability of AI to exceed humans. But I sure like going to Google rather than spending hours in a library and then not finding what I'm after. Unfortunately, .com and business have taken over search engines. So, finding research articles from .edu or .org becomes harder with all of the commercialism in most searches. AI helps researching well, but analyzing, serendipity and synthesizing is something that will take a long time for AI to do well enough to compete with human minds.
Ron