On IBM Researching Artificial General Intelligence
That artificial general intelligence, the ability to understand and learn an intelligent task humans can understand and learn should be research that you work with in the lab. … I don’t want us to work only on moonshots. We have to work on those things that are near end, that provide value this year and next year, in the year after, for our clients.
In addition, we might work on a couple of what we are labeling as ‘moonshots.’ Look, generalized artificial intelligence, personally, I think is still a long time away.
I think the last time we polled scientists—not just IBMers but other scientists—you get a spread. Some people believe it could be as early as 2030. But the vast majority put it out in the 2050 to 2075 range.
Personally, having grown up as a scientist, I react with anything that is 25 years or further away—my conclusion is that means we have no idea how this is going to happen.
So is it worth working on? Sure. Is it worth making that the majority of the effort? I think it is too stretched.
