Science Blasts Past Science Fiction

by James Wallace Harris, May 14, 2024

Since the 1960s I’ve fantasized about writing a science fiction novel. Since the 1990s, I’ve fantasized about writing a science fiction novel about artificial intelligence. The likelihood of me ever writing this novel is nil, but I still imagine and refine it in my head. The trouble is, artificial intelligence has gotten real, making writing a novel about AI more difficult.

Yesterday OpenAI announced ChatGPT 4o. (That’s a little o – pronounced four oh.) If you haven’t seen any of the news stories or demoes, watch this video.

I first encountered the concept of artificial intelligent back in 1967 when I read The Moon is a Harsh Mistress by Robert A. Heinlein. That novel featured a sentient mainframe computer named Mike. He was my favorite character. Heinlein didn’t use the term AI. I wouldn’t encounter it for a few more years when I went to a two-year technical school to study computers. Along the way I read two other books that featured a sentient computer, When H.A.R.L.I.E. Was One (1972) by David Gerrold and Galatea 2.2 (1995) by Richard Powers. There were plenty of other novels and movies about intelligent computers, but these three were my favorites.

In other words, I’ve been waiting a long time for science to catch up with science fiction. It’s funny though, when reality finally caught up and surpassed fiction, it’s been kind of anticlimactic. Most AI science fiction today has more in common with 1960s science fiction than it does with 2020s computer science. I’m currently reading a 1983 story by Poul Anderson, “Vulcan’s Forge” that has a sentient computer, but not much different from Mike in Heinlein’s 1967 story, or the female AI in the 2013 film Her. It seems now that science fiction writers never pushed their imaginations enough.

ChatGPT 4o is way cooler than anything I’ve ever encountered in science fiction. For one thing, I’ve always assumed an AI computer would talk like a person, capable of holding only one conversation at a time. Real AIs now converse like real people, but with thousands, if not millions of people at once. Never imagined that. And ChatGPT 4o sounds like a real person. I’ve always assumed they would sound like a computer.

I also thought intelligent AIs would have to be sentient beings. Evidently, massive amounts of intelligence can still be unaware even while coming across like people. We’ve always assumed that the growth of intelligence paralleled the growth of consciousness. That might not be true. Of course, there are philosophers who claim that consciousness is an illusion. Maybe they were right. But where does that leave us?

Computer scientists are quick to point out that ChatGPT 4o isn’t real AGI (artificial general intelligence). Critics constantly point to the factual mistakes that even the best of current AIs make. The old Turing Test concept only required a human not to be able to tell the difference between a machine and human when communicating through a teletype. Humans make factual mistakes all the time. Both presidential candidates frequently misspeak, but we don’t accuse them of hallucinating. That’s the term computer scientists use when AIs make up shit. If we applied the same term to humans, humans hallucinate their asses off. I’m not sure the average person can tell the difference between a computer and a person anymore in a Turing test. And neither can we tell the difference between AI and AGI.

For many measurable areas of information, current AIs are far smarter than humans, and are getting smarter all the time. They may not be perfect at any single discipline, but they are far better than average at most subjects they are tested on. When computer scientists talk of AGI, I get the feeling they are expecting the machines to know absolutely all about everything. Haven’t they up the ante to AO (artificial omniscience)?

Here’s the thing. I always thought the world would go nuts when AI singularity happened. And it may have happened or may not. But that’s my point. I think it has happened and we’re just not smart enough to recognize it — yet.

People are embracing artificial intelligence in the same way they embraced smartphones and the internet. They just incorporate recent technology in whatever way they find useful or advantageous. They don’t philosophize about it. They might bitch they’re losing their jobs, but not because AI is a science fiction sense of wonder come true.

Why did talking computers evoke a sense of wonder in science fiction, but not in real life? How come I haven’t seen one person’s mind blown? Not even a, “Dang! I’m talking to a dang computer!” Instead, people are whining about AIs making math mistakes or getting dates wrong. Every human I know does that, so why not cut the machines some slack?

Why isn’t the current state of AI development a boom time for philosophers and science fiction writers? People treat ChatGPT 4o like an iPhone 15, just enother gadget. The only speculation seems to be when will we get ChatGPT 5 or the iPhone 16?

JWH