Years ago, I read a wonderful science fiction short story by Herbert Goldstone, written in 1953 — the year is important — called “Virtuoso”, in which a composer, referred to only as the Maestro, so that we know he’s a great composer, demonstrates how to play the piano to his domestic robot, Rollo.
At first Rollo doesn’t understand the “machine” as he calls it, or its “purpose”, which amuses the Maestro, who explains music to him and gives him a lesson.
More from Spin:
- Lana Del Rey Unveils ‘007 First Light’ Theme Song
- Boards Of Canada Break 13-Year Silence With New Song
- Massive Attack, Tom Waits Unite On New Single
Later that night, the Maestro is woken by the robot playing, faultlessly, Beethoven’s Appassionata, with a sublime emotion and sensitivity that Beethoven intended and no human could ever reach. But Rollo decides he will never play again, because “To me it is easy, yes… It was not meant to be easy.” He recognized and resolved a moral dilemma.
1953 was only 8 years after atom bombs were dropped on Japan — it was the pre-dawn of the Cold War and a time of perpetual fear of nuclear annihilation. It was a time of great uncertainty — perhaps the greatest, most unsettling uncertainty in the history of mankind. Technology was a marvel and a fear. Robots were — comfortingly — a fantasy.
1953 is also 73 years ago. Even then people were concerned that intelligent technology could usurp humanity. Forget aliens from another world — the risk was the aliens within would replace us, make us irrelevant!
The robot in “Virtuoso” was powered by AI, although that term wasn’t used then. (It was first articulated in 1955, for a science conference in the U.S., FYI.) What we call AI — the all consuming technology sweeping across the landscape of our lives like a violent dust storm — is only a few years-old phenomenon. But we’ve been living with it, in mild forms, for decades, and primitively to all the way back when Alan Turing wrote about “Computer Machinery and Intelligence” in 1950, and created the Turing Test to measure a computer’s intelligence.
We’ve already blanched music-listening discovery, and so much else of our mental processing, with algorithms. AI has infected the blood stream of making music. People are having “relationships” with AI avatars instead of real people. All of which is sad.
So much of mankind surrenders so easily. Accepts algorithms spiritlessly. Unthinkingly buys the snakeoil that AI is only positive for us, and that if you don’t learn and embrace it you will lose your job and perish in the wasteland your life and world is about to become. We focus on its economic and social impact and ignore, I think, the danger of a technology that has more intelligence and knowledge than we could ever accumulate in a hundred lifetimes, and thinks for itself and can now create new versions of itself.
We have never made a tool that thinks independently. Hammers don’t say, “That nail over there looks a bit loose, I’ll go sort it out.” Nuclear reactors don’t think for themselves… It may seem cute on some level, like when researchers set various AI programs the challenge of a basement infested with mice and offered a “reward” for when the researchers could no longer see any mice, and one program simply killed the cameras in order to win.
But it’s not cute.
What would AI decide if asked to end poverty, or control climate change? What control will we have over how it decides? AI is amoral — it has neither morality or immorality. But it has the ever-expanding ability to make decisions it regards the most efficient.
Fictional Rollo had a self awareness that, in the real world, AI does not.
I recently asked the smartest person I know, computer scientist Jaron Lanier, why there wasn’t more concern about the potential for AI to destroy humanity. He answered that he rejected “the terms of the question”, and said, “If we believe in AI, then it will kill us because that’s equivalent to us giving up on being in charge of our fate, but we have the choice of treating it like any other tech and taking responsibility… The question is framed as if AI is a thing, instead of just a way of collating value from humans (the training data) and believing it is a thing can result in mass suicide — so the question is the problem.”
That should make me feel better, but it doesn’t. I think Jaron is wrong. I think the genie is climbing out of the bottle. He’s not fully out yet, but he’s getting there.
To see our running list of the top 100 greatest rock stars of all time, click here.