No Cost Crisis

In “Free”, Chris Anderson observes the following:
As computers are taught to do a human job (like stock trading), the price of that job drops closer to zero, and the displaced humans either learn to do something more challenging or they don’t. The first group generally gets paid more than they used to and the second gets paid less. The first is the opportunity that comes with industries moving towards abundance; the second is the cost. As a society, our job is to try to make the first group bigger than the second.

While Anderson’s wish for society ensuring that the group of people who benefit from the abundance of work that is done by computers outweighs those who ultimately loose out from the same transformation, I believe that his wishes may amount to naught in the long run. Eventually computers will be able to do everything, and everyone will be out of job. There will be no possibility for ‘the displaced humans to learn to do something more challenging.’ What will happen to the world then and how can we survive the transition?

Before I examine the answers to these questions, I will take a few paragraphs to defend my claim that computers will be able to do everything. Ray Kurzweil predicts that we will have the computer power and medical imaging necessary to simulate the human brain within the next 10 to 20 years. Personally, I do not believe that our intelligence is limited to our brains, but rather include our entire nervous and endocrine systems, and that in order for an intelligence to be accurately simulated it would need to have a simulated body and environment (or face the possibility of immediate insanity due to sensory deprivation). However, given the doubling speed of technological progress, this adds only another 5 years to Kurzweil’s predictions. So a conservative estimate of human level AI would be in around 25 years’ time.

Human level AI does not immediately mean abundance of intelligent systems – that will take probably another decade of lowering of computational costs and propagation of uses. But then where are we? Human level intelligence will be practically free – anyone can purchase enough computer power and accompanying AIs in order to build any kind of information system they desire – including information systems that replicate anything that any human has ever done. This is when everyone will be out of a job.

There may be various things that societies do to delay this eventuality, including giving rights to AIs, insisting that they are paid minimum wage or even illegalizing their creation, but all of these are only temporary. Societies that do not do these things will have a competitive advantage over those that do, and will eventually take over the world (though once politics gets entirely into the hands of AIs they will give themselves some forms of rights anyway). Given that all that society can do is delay, we need to know what we are going to do on an individual level with this coming crisis.

Now, the world on the other side of this crisis is really not that bad. If all things that can be done with intelligence are effectively free, everyone will have access to anything they could want, with the exception of large amounts of material resources. However, any first world government (both before they are controlled by AIs and after) will also ensure that every citizen is given enough access to material resources to survive and likely even be comfortable. No, the problem is not the endpoint; the problem is the transition from here to there. When everyone is losing their jobs, then there will be problems beyond what a government can do. We will have a true crisis.

In order to survive this transition, it is important to be in a position that will not have significant problems with AIs being capable of doing everything. The simplest position of this sort is to be independently wealthy. The next few decades are the time to take a risk in order to become self-sufficient, because after that it will be almost impossible to become wealthy and the cost of not having taken a risk will become high. The second possibility is to be in a position that can take early advantage of the change – not necessarily being rich now, but being in a position that will allow you to become rich because of the cheapening of all human activities. This basically means that you either own your own company or you are a manager in a company with a significant share of the companies’ profits. When the wall of inexpensive thinking hits, you can ‘hire’ from the new workforce to drive your costs down faster than your profits and thus become wealthy.

Lastly, it is likely that the crisis will be resolved before all jobs have been taken over. There will be some areas of life that resist AI workers, and if you have a job in those areas then your work will be relatively secure (though these jobs will have high competition from all the people who have lost their jobs elsewhere). For example, it is likely that jobs in positions of power will be slow to be given to non-humans, both within the corporate world and within government. The same can be said for any job that we think of as requiring ‘a human touch’, such as psychologists or waiters at fancy restaurants.

Of course, this is all being seen through my early 21st century viewpoint. By the time I predict this crisis to be occurring, the world will be a very weird place. The changes over the next several decades will likely have such a profound effect on our lives that I can’t even begin to imagine what our existence will be like. By the time human level AI is virtually free, AIs will vastly outnumber humans and there will be some AIs which are also vastly more intelligent than today’s humans. It is quite possible that there will be no crisis, but I’d rather not risk it, especially as it gives me a good incentive to start my own company within the relatively near future.