Edge Cases
I’ve been reading a lot about the history of automation and resulting job displacement. I think it’s really important to understand how things have gone before so that we are best positioned to manage transitions appropriately going forwards.
I am obviously not against the progress and deployment of technology. It would be impossible to be so and also work in Venture Capital: the two are incompatible. It’s also naive to push back against it in a globally competitive economy: if we don't get ahead of it, we’ll simply lose ground (i.e. economic might) to those who do.
But I do think we need to be conscious of the social and economic disruption potential that AI could bring. It’s easy for the tech-savvy utopians to extol the virtues of the advance of software over labour: they stand to benefit, after all. But we need to think through how the rest of society manages the change.
One useful framework might be to maximise our human excellence at the edges. Computer models are great at predicting probabilistic outcomes, but humans are masters of the edge case. That’s one of the reasons technical problems like self-driving cars and general robotics are such hard problems: catering to the edge case is definitionally difficult because of the lack of data to train on.
Humans, by contrast, have ingenuity and initiative that they can deploy at will. Being absolutely brilliant at edge cases might be safest place to be. In other words: avoid the repetitive and predictable, and embrace the unique and volatile. The former is the purview of machines. The latter, of humans.