Photo by Aideal Hwa on Unsplash
Now that everyone is talking about ChatGPT, I’m finding myself dodging some version of this question more often:
“Will AI fully exit the realm of science fiction and begin to change everything?”
The reason I usually nope out of this question is that there’s some sneaky sleight of hand in there which ensures that intelligent discussion will be dead on arrival. That’s because language is a slippery eel and there’s no law against people using the same word to mean different things in different contexts, but let me indulge the question just this once.
There’s no law against people using the same word to mean different things in different contexts. That’s what’s happening when people talk about AI these days.
Let’s look at some of the ways that people are using the term AI to talk past one another. When we see those for what they are, the answer to our slippery question will slither into the light of day all by itself.
Homonym 1: The AI made up by writers
The AI you’ll find in the pages of science fiction novels is entirely made up! Most of the time, it plays the same role in the narrative as all the other conveniently creepy almost-but-not-quite-human entities — demons, clones, demigods, aliens, golems, spirits, talking animals, animated puppets — that force us to confront what it means to be human. I hope no one is surprised that the sentient human-adjacent evil antagonist version of AI doesn’t exist.
Hate to disappoint, but a murderous HAL 9000 isn’t going to be piloting your spaceship anytime soon.
Science fiction is simply called “fiction” when it doesn’t stay ahead of technology, so in a breathtaking display of how tautological tautologies can be, let’s get this version of the question out of the way: will every science fiction version of AI fully exit science fiction? No. Science fiction will keep writing about new tomorrows.
The sentient human-adjacent evil antagonist version of AI doesn’t exist.
I’m that special type of curmudgeon who thinks it’s a crime against intelligence to talk about machine “intelligence”, “sentience”, “consciousness”, and “the singularity” without first defining clearly and unambiguously what these words mean to you. It’s not cute to work yourself (and your audience) into a frenzy by saying a whole lot of nothing, but unfortunately that’s what so many discussions on these topics turn into.
As for making artificial humans with all our desires, feelings, and foibles… why would anyone even want to do that?
As for making artificial humans with all our desires, feelings, and foibles… why would anyone even want to do that? If you stop to think about it, that smells so much more like a plot device in a novel than like a serious goal. Regardless of whether you’re a bleeding heart or have no heart at all, if you wanted machines that could be most useful to you and/or humanity, you’d opt to build them without selfishness and other less lovable aspects of human nature. Whereas if you simply need more humans, buy me a beer so I can explain exactly how they’re made.
Image source: Pixabay
To be sure, there are great reasons to worry about AI, but the it-reminds-me-of-me uncanny valley stuff is a distraction. Since I’d need a separate whole blog post to explain this, I’ve… written a separate blog post for you here.
Homonym 2: The AI people like me are talking about
In tech, we use the term “AI” to refer to a specific way of turning data into computer code. When you see someone automate a task using patterns in data without looking up the answer directly, they’re probably using machine learning to solve their problem. If it’s a particular type of machine learning algorithm, then it’s polite to call it AI, though etiquette never stopped certain “AI” startups from jumping on the bandwagon with their three IF statements in a trenchcoat. (You know who you are.)
If you'd like to meet the real AI, check out this excerpt from my machine learning course on YouTube.
So, has this version of AI already exited the realm of science fiction? Absolutely, since it never even entered it. Even before all the AI applications we use every day were launched, the people doing serious work on them were calling them AI, fully aware that they weren’t creating anything that would pass muster with either science fiction fans or professors of neuroscience. It’s just a flashy name for some math. Aren’t homonyms fun?
But as it turns out, that math is incredibly useful! It enables all kinds of interesting applications, and you interact with them every day. They’ve been “changing everything” for the last decade, from our smartphones and laptops to our cars and homes. This kind of AI is quiet and unassuming, and it makes our lives more convenient without drawing attention to itself.
Every time I see someone “fail” a Turing test, it tells me a great deal more about the person than about the machine.
And that same math happens to be useful for processing audiovisual data so it’s a solid ingredient for all those parlor tricks you’d need to fool the gullible into thinking you’ve built a sentient being. But every time I see someone “fail” a Turing test, it tells me a great deal more about the person than about the machine.
Just because telekinesis looks real to you doesn’t mean magic is real. Just because you can’t tell a machine’s not a human doesn’t mean it is sentient. I'm perpetually surprised by a world in which professional magicians and the Turing test can coexist. Photo by Fengyou Wan on Unsplash
Finally, I’d be remiss if I didn’t mention that there are two more homonyms nested inside the one in this section — the AI that’s more associated with academia and the AI that’s more associated with industry. If you’re curious about the difference between two more homonyms, check out my explainer in “Why Businesses Fail At AI.”
Questions to sink your teeth into
“Will AI fully exit the realm of science fiction and begin to change everything?” The answer is tautological either way. You’ll answer no by simply reminding yourself of any reasonable definition of science fiction and you’ll also answer no because applied AI hasn’t had all that much to do with the science fiction version for at least this millennium.
There’s no more controversy here than two groups talking at cross purposes without defining their terms. Instead, let me try to extract some properly controversial questions from this homonym soup:
- Will the popularity of practical AI kill science fiction’s AI? I could see an argument for the term becoming too mundane to be compelling… perhaps one day an AI antagonist will be as hard to take seriously as a disgruntled toaster. Will science fiction writers need to look for other entities to fill their uncanny valleys? My guess is no — if creepy dolls are good enough for the fiction community, then apparently anything goes — but what do you think?
- Did the builders of practical AI solutions do the right thing by allowing their marketing teams to capitalize on science fiction hype in order to secure funding and gain attention? How different would technological progress have been if we stuck to terms like “layered weighted sums of nonlinear data transformations” instead of flashy phrases like “artificial intelligence” and “neural networks”?
- Should we let nature take its course and let the public continue to confuse practical AI with sci-fi AI, or do we need to actively educate people about the differences between the two? If you’re interested in my thoughts on why it’s dangerous for society to conflate the two, check out my blog post titled “Forget the Robots, Here’s How AI Will Get You.”
Excited to hear your thoughts as you digest these chewy questions — join the discussion here.
Thanks for reading! How about an AI course?
If you had fun here and you’re looking for an applied AI course designed to be fun for beginners and experts alike, here’s one I made for your amusement:
Enjoy the entire course playlist here: bit.ly/machinefriend
P.S. Have you ever tried hitting the clap button here on Medium more than once to see what happens? ❤️
Looking for hands-on ML/AI tutorials?
Here are some of my favorite 10 minute walkthroughs:
/w=3840,quality=90,fit=scale-down)
