AI of itself is deeply concerning. In theory, it could generate enormous benefits (for whom is the thorny question here though) along with many unintended consequences. For those of you who like to point out the fear of technology is nothing new, I would like to point out this time it’s utterly different.
With previous technology, the fall-out was always human: loss of jobs, lower wages, retraining, etc. That was because the technology was entirely predictable. A robot, for example, is designed to be more efficient at doing certain jobs. But it has to be told what to do. It’s dumb.
AI is different because its purpose is to be intelligent. It’s designed to be adaptable under different circumstances. Machine learning algorithms are supposed to generate innovative solutions. That means AI must be unpredictable if it is to be successful.
That’s concerning enough but here is where it gets positively scary.
AI is not one thing that benign scientists are collaboratively trying to coax into life through careful steps for the benefit of all humanity. Oh no, AI is a pristine gold seam a mile wide and rapacious multinationals are fighting to get in there first. That means there will be many AIs operating in a market place and a capitalist market place is a thoroughly ruthless place.
You may recall the emissions test scandal for diesel cars that Volkswagen perpetrated (although other companies were at it too). The purpose of the test is to reduce harmful emissions. You would therefore imagine engineers would try to work towards this goal because everyone benefits. But no, someone in the VW organisation (remember, in any given group, it’s the top executives who harbour the greatest psychopathic traits) decided the test was to be gamed. By gaming the test, only the company would benefit and more cars could be sold. Multinationals are only interested in one thing — money.
You see the problem now? These same multinationals are all racing to get their AIs into the market. If it’s a tight race, they will be tempted to game the system and take risks. AI is already a huge risk. Allowing psychopaths to be responsible for some of it and have them tinker with the outcomes is a guaranteed path to destruction.
I can almost hear the voice of the gangster executive from the latest AI company introducing their latest product: say hello to my leetle fren…