Google’s presentation about AI assistants calling hairdressers and restaurants rattled the news. The conversations definitely seemed to pass a Turing-test, raising moral and ethical questions about the use of AI. Google says that such AI interactions will inform the other end about the nature of the call, so we have little to worry about. Beyond the arguments whether the spread of AI will lead to a dystopian future, where James Cameron’s Terminator or the Wachowskis’ Matrix will become reality, I found the economic argument just as threatening.
The rise of internet based economy, robotization, and machine learning already turned most economic sectors, thus most people’s lives upside down. There were no need for machines turning on humans, we did it ourselves, mainly through protest votes and sometimes by putting incompetent populists into leadership positions. How governments, decision makers, and the market will regulate AI will be more interesting to follow than speculating on how to defeat SkyNet.
Firstly, let’s look at a simple example: who was the main benefactor of industrial revolution in the UK? The technological advancements definitely increased profits of the clothing industry, but mainly through economies of scale, lowered costs, and thinner margins on each product. The more compelling answer is that, as always in economics, the coin has two sides. The clothing industry definitely benefited from the implementation of new production methods, but so did consumers who could buy affordable clothing. Who are the losers? Inefficient competitors and a portion of the workers.
Secondly, let’s decouple the phenomenon with economic terms. As I always say, get rid of the money factor in such analysis. What we end up with is a part of the economy, who produces clothing, and another part which produces something else and demands clothes. They exchange at a given rate, determined by relative prices. If suddenly I can produce more clothes with the same effort I will have extra purchasing power, extra goods I can exchange. How is this extra welfare captured?
If the other side of the economy has no demand, the extra value will be zero for both of us. If they have demand, but no spare capacity then I can only outbid someone else with my extra clothes, and receive the same amount of goods. The outbid people might not get what they originally wanted, but my vendor captured the entire extra value, which she can exchange yet for something else. As you can see in the latter case the value is actually not captured by the innovative producer (me in the example), but by my consumer. However, my consumer can once again use the extra purchasing power, which can be passed on until it finds extra capacity somewhere. On an aggregate basis it sounds attractive. Yet this path restructures the economy which always carries the risk of turbulence, or market failure in extreme cases. The last possibility of value capturing is the most desirable and simple. I have extra clothes, my consumer/vendor of other good has spare capacity. Both of us end up with more goods what we desire.
Thirdly, have a look at today’s case. The rise of AI, just as any other efficiency improving technology, will create extra value in industries where they are implemented. If a wide segment of the economy is able to leverage on it simultaneously, that is the closest we can get to the most desirable scenario.
Finally, we must talk about redundancy and its effects on labour. Once again, if the broad segment of the economy is affected by the change the demand will increase alongside the capacity. Thus there will be no need to sack people. However, this has one condition. People must be able to produce by using the new technology. The biggest threat to implementing AI is not its visioned rise against humanity, but the uneven development and the faith of unskilled labour.