Humans Need Not Apply – Jerry Kaplan

I’d started reading Humans Need Not Apply on NetGallery, but ran out of time before it was due back, so ordered myself a paper copy. It is a fascinating perspective on what Artificial Intelligence (AI) will mean at the individual level. This is not only the “you won’t have to drive your car” but also “your job can almost certainly be automated”.

Consider a robotic housepainter. It’s easy to imagine a humanoid form climbing ladders and swinging a brush alongside its mortal coworkers. But it’s more likely to appear (for instance) as a squadron of flying drones, each outfitted with a spray nozzle and trailing a bag of paint.

That is just an example of the sort of thought experiment that Kaplan uses to illustrate this world, where humans become less necessary every day, but then how we would change to cope with this world, and find our place in it. He explores the moral implications of creating and owning AIs, and how their legal status could change over time. As a long-time reader of science-fiction, I find this speculation fascinating. including the Questionable Content (usually safe for work, despite name) webcomic, which has rights for AI as a major theme.

I’d also read this straight from a book on how to efficiently externalise effort, and another on the moral implications of new tech. It made for an interesting combination.


Bad maths time…

The chance that someone random will click on an ad for a golf vacation may be one in ten thousand, but if you are male, it may increase to one in a thousand…

Kaplan let himself down badly here – as LevitinĀ remarked, very few people understand Bayes’ Theorem

Start by taking the original 10,000 people. One of these will click on the ad.

Assume no woman would, and that women are (more or less) 50% of the population.

There are now 5,000 men, one of whom will click on the ad, i.e a one in five thousand chance, and Kaplan has lost 4,000 men. The two statements in the quoted sentence are not logically compatible.


I’m also not convinced by the statement (on self-driving cars) that…

Traffic jams will be a quaint memory of more primitive times.

As the main problem here is not lack of coordination but simply capacity. Unless the swarm is programmed to not enter congested areas, in which case the problem will not be traffic jams, but holding pens.

 

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.