Avengers, the Age of Ultron opens in the UK later this week, and in the US the week after. Apparently Hollywood can forecast a film’s takings pretty well these days (thanks to clever AI algorithms, no doubt) and it seems the studio is quietly confident it’s going to overturn box office records.
It may also overturn something else: the unwritten law that every article about the future of artificial intelligence has to be accompanied by a picture of Arnold Schwarzenegger, or the killer robot he played. The original Terminator movie was released in 1984, and 31 years is a great innings. Now Ultron the Marvel super-baddy may well become the new iconic image of a rogue AI.
Ultron, an AI created by Iron Man, has been a key character in Marvel comics for years, but he is new to movie audiences. He arrives at an interesting time: during the Terminator’s tenure, the public has not taken seriously the idea that intelligent machines could become a threat to mankind. Now, thanks to the amazing advances in weak artificial intelligence, and specifically due to the publication last year of Nick Bostrom’s book Superintelligence, that idea is treated with less disdain.
The growing awareness that AI is an increasingly powerful technology which can have negative as well as positive consequences is to be welcomed. But there is a danger that only the negatives will be understood. In a world of short attention spans where good news is no news, there is a danger of an ill-considered backlash against AI research. That could deprive us of the tremendous benefits that AI can bring us in the years ahead, and in the worst case could drive AGI research underground, to the detriment of the Friendly AI work which we need to accompany it.
Science fiction books and movies provide us with vivid metaphors – shorthand for future scenarios. To have the nuanced, thoughtful debate that we need to have about AI we need a wider range of metaphors. Ultron should provide lots of fun, but nuance…? Probably not so much.