Wednesday, January 07, 2015

Predictions

Prediction is difficult, especially if its about the future...............Niels Bohr.

Nonetheless, nature equipped us with these large brains, and they are useful mainly to the degree that they help us predict the future. Some kinds of predictions are safer than others - especially those solidly based in physics and experience. Some of the really hard ones involve politics and economics. That's mostly because those subjects are complicated and subject to large perturbations by unpredictable events.

Our predictions are all based on the idea that the future will be analogous to the past. Computers will continue to increase in power is a prediction made many decades ago and it seems certain to continue for at least a while. Sensors will continue to get better, cheaper, and more ubiquitous is a similar durable prediction. Artificial intelligence will continue to become more capable and incorporated in more and more devices.

This last prediction may be more controversial, but mostly only among those who really haven't been paying attention. It's essentially an inevitable consequence of the previous two predictions in combination with with the rapid progress being made in animal cognition.

In the past couple of decades, artificially intelligent machines have shown that they can do a number of tasks formerly considered intellectually difficult (in chess, pathology, and law, for example) better than humans. The next generation of AI will produce devices that are increasingly autonomous - capable of doing various tasks humans do without human supervision, or with minimal human supervision. Typically these tasks, from bagging groceries to shooting down fighter planes, will be tasks that formerly required not vast intellectual power but an integration of sensing and decision making of the type that now or formerly required a person.

The economic consequences of these developments have been depressing wages and employment for decades now. It seems likely that these effects will continue and be exacerbated in the future.

Nouriel Roubini, AKA Dr. Doom ever since he predicted the Great Recession, has some thoughts on technology, including some of the downsides that I see. A sample:

In my view, from the economic perspective, the technological forces driving this revolution tend to have the following three downside biases. That is, advances in technology tend to be:

capital intensive (favors those who already have money and other resources);

skills biased (favors those who already have a high level of technical skill); and

labor saving (reduces the total number of jobs in the economy).

The risk is that workers in high-skilled, blue-collar manufacturing jobs will be displaced by machines before the dust settles at the end of the Third Industrial Revolution. We may be heading toward a future where factories consist of one highly skilled engineer running hundreds of machines—with one worker left sweeping the floor.

In fact, the person who sweeps the floor may soon lose that job to a faster, better, cheaper, industrial strength Roomba Robot!

It's number 1 I worry about. The benefits of the productivity gains of the last three decades have gone overwhelmingly to the top 0.1% and especially to the top 0.01% This is not a recipe for a stable society.