First let me just say the term artificial intelligence is nonsense and will have to change in the near future. There’s nothing artificial about Digital Intelligence (DI).
But I digress before I’ve even gotten started.
In the Near Future, DI will be essentially indistinguishable from our own. There are lots of very smart and well-funded people working very hard on this. The race is on, and they have already come a long way.
The Emerging new systems (Google Large Language Models) “…can generate paragraphs of human-like text based on what it’s learned from a vast database of digital books and online writings.” Matt O’Brien recently wrote in AP news. He also said they, “… can converse, generate readable text on demand and even produce novel images and video.”
Add these emerging digital language skills to the equally fast growth in human like robotics and we would be wise to envision a near future in which the knowledgeable professional sitting across the desk from us will not be a person.
One other equally important consideration must be remembered. There will remain a subtle difference. Let’s take a look at the current flaws in these systems as a method of illuminating their future imperfections, which may go otherwise unnoticed.
Matt O’Brien reported on one of the top digital language systems, GPT-3. “GPT-3 can write up most any text you ask for — a cover letter for a zookeeping job, say, or a Shakespearean-style sonnet set on Mars. But when Pomona College professor Gary Smith asked it a simple but nonsensical question about walking upstairs, GPT-3 muffed it.”
“Yes, it is safe to walk upstairs on your hands if you wash them first,” the AI replied.
Sorry GPT-3! That’s wrong. When I walk upstairs on my hands, I don’t wash them until afterwards.
Matt O’Brien wrote, “As AI systems are increasingly able to write health advice websites, high school term papers or political screeds, misinformation can proliferate and it will get harder to know what’s coming from a human or a computer.” I would correct Matt slightly. It will not “get harder”. It will become virtually impossible.
On the lighter side one blogger, Janelle Shane, author of the AI Weirdness blog, has creatively tested GPT-3 — The results are interesting and amusing.
Issue 1. In your own blog you began with “…let me just say…”. Perhaps you feel the need for permission from a nonexistent authority, and if so, give yourself permission to put that feeling in a jar and quickly close and tighten the lid, then bury it.
Issue 2. You allege that “…the term artificial intelligence is nonsense and will have to change…” and yet the widely accepted definition from bing-oxford is “made or produced by human beings rather than occurring naturally, especially as a copy of something natural.” This definition makes complete sense, not nonsense. AI is made by humans as a greatly improved copy of some of our natural brain functions. Yes, AI is digital.
You have a history in this blog from Eight Parties to little b bigotry, of misdirected redefinitionism. Of course, it is true that some accepted terms are wrong. But I’ll be glad to help, contact me when you get an urge to redefine, or blog it.
Issue 3. No need to apologize to GPT-3, we all make mistakes. When AI Watson crushed top champions Ken Jennings and Brad Rutter in Jeopardy 11.5 years ago, Watson made some mistakes. None of that was your fault and if you feel that way, repeat the jar and burial advice given above.
Stigman Roid, Psychobabblist, I accept your professional assessment with glee.
As to issue 2: I accept bing-oxford in general. It is a very good definition of AI as definitions go. Just to be clear, my argument is not with the “intelligence” part, but with the “artificial” part. You are right to say that AI is “..a greatly improved copy of some of our natural brain functions.” And the fact that it is a “copy” does imply artificiality, but many of the other connotations of the word artificial do not fit. AI is not dead or lifeless like an artificial plant for instance. “Lifeless” is problematic. Let’s say “Inert.”
Thanks Stigman for my new title, misdirected redefinitionist. I will wear it proudly and do my best to live up to everything my new title infers about me.
In Issue 3, you mentioned that Big Blue made mistakes. That will rapidly become less frequent and much, much harder to detect! I am adding a Youtube video I think you and everyone else should find interesting. It shows some of the recent work being done in AI by a group calling itself Brilliant (and rightly so).
Click on the word Link.