I just read an article by Ezra Klein in this week's Sunday NY Times on Artifical Intelligence which makes me feel as if my friend Bill and I have been busy shifting the chairs on the Titanic.
He and I have been fussing over the way we've been duped into thinking we're reading true stories on the internet, only to discover at some point that they are AI generated. The solution to our complaint, obviously, is to seek legislation requiring YouTube (and other publishers) to state explicitly and openly whenever what they publish is AI generated. Easier said than done, of course, but that's a fight for another day.
This Ezra Klein Times article (Sunday, September 20) points out that a far more dangerous concern is what is known as RSI - "recursive self-improvement." That is, the ability of artificial intelligence to ignore human input entirely by getting around its limitations - by combining efforts with other AI generators, for example - and generating ever new and improved ways of producing information.
There is no way of preventing a mechanical (non-human) producer of "truth" from producing things that could lead to the taking of human life as a useful shortcut to efficiency or productivity, for example, unless we have some sort of human guardrails, some form of ethical constraint. If left to its own devices, AI programs have no "reason" to select a moral constraint - or any other constraint. As non-humans, they have no morality. Just as corporations lacking CEOs with a conscience have no goals other than profit, AI programs, which lack a sense of ethics, have no goal other than a solution to any given problem it is working on. Damn the torpedoes.
A sense of morality can, of course, be programmed in. But the way the world is currently set up, with nations in competition with each other, no one nation is going to place such constraints on AI for fear that it will allow other nations to soar ahead in their own particular AI development. And this is true not just for competitions between nations, but for competitions within any given nation as well.
The way that nuclear proliferation got out of control should be a warning that this threat is not merely theoretical. It's real. And of a much greater degree of magnitude.
Makes me glad I won't have to live in a world where this could happen.
Unless my doctor's warning that instead of dying within the next couple years from pulmonary fibrosis I could live to 90 and be hit by a bus is unduly pessimistic and I will live to 100.
No comments:
Post a Comment