“And pray that there’s intelligent life somewhere out in space, ’cause there’s bugger all down here on Earth.” —
“The Galaxy Song” (“Monty Python’s The Meaning of Life” – 1983).
– – –
It’s very popular to trash “artificial intelligence” (AI) these days.
While reasoned warnings regarding how AI-based systems could be abused (and/or generate inappropriately “biased” decisions) are appropriate, various folks in the public eye — some of whom really should know better — have been proclaiming nightmare scenarios of AI relegating we mere humans to the status of pets, slaves, or worse — perhaps “batteries” as in “The Matrix” (1999). Or maybe just fertilizer for decorative displays.
There’s certainly a long history of cinematic representations of “intelligent” systems run amok. Earlier than “The Matrix” we saw a computer lethally hijack a space mission (“Hal” in “2001: A Space Odyssey” – 1968); another computer imprison, rape, and impregnate a woman (“Proteus” in “Demon Seed” – 1977); and a pair of computing systems take over the world (“Colossus” and “Guardian” in “Colossus: The Forbin Project” – 1970). And of course there’s the scorched Earth world of “Skynet” in “Terminator” (1984), and hybrid threats that may be even scarier, like the “Borg” from the “Star Trek” universe. And more, many more.
All of these cultural references have a real impact on how we think about AI today. We’re predisposed to be fearful of systems that we believe might be “smarter” in some ways than we are.
Monty Python may have been partly correct all along. In a world where a moronic creature like Donald Trump can be elected to the most powerful role on the planet, we should probably be seeking out intelligence to augment our own, wherever we can find it.
Seriously, since it could be a long, long time (if ever) before we hear from interstellar civilizations (and Stephen Hawking’s prediction that this might be a seriously losing proposition for humanity could indeed be accurate), we need to concentrate on intelligence augmentation systems that we can build ourselves.
The word “augmentation” is crucial here. The human brain is a marvel in so many creative and imaginative ways. But it’s easily overwhelmed by data, subject to disruptive distractions, and is ill-suited to solving critical planetary-scale problems on its own.
The key to a happy coexistence between humans and AI systems — even advanced AI systems — is to keep in sharp focus where we excel and where the AI systems that we develop can be most effectively and successfully deployed.
Two ways that we can get into trouble are by trying to use AI and “expert systems” as shortcuts to solve problems for which they aren’t actually suited, or by assuming that the data that we provide to these systems is always accurate and fair, when in some cases it’s actually biased and unfair (we’ve seen this problem already occur in some systems that attempt to predict criminal recidivism, for example). The computing adage “garbage in, garbage out” is still true today, just as it was in the ancient era of punched card computing.
Obviously, we don’t want to screw this up. There are real challenges and significant (but fascinating!) issues to be solved in the ongoing development and deployment of AI systems going forward, and in helping non-technical persons to better understand what these systems are really about and how they could actually improve their lives for the better.
And to the extent that we can concentrate on the real world of AI — and less on dramatic “doom and gloom” scenarios straight out of the movies — I believe that we’ll all be better off.