Why Machine Learning Matters to Humans

The cultural consequences of AI

November 11, 2016 - 3 minute read -
cs machinelearning

The field of statistical machine learning has been around for a long time, but it has always been limited in its ability to do simple tasks that humans do with ease every day. You don’t see it as AI, but you are almost guaranteed to be exposed to these classical methods on a daily basis in the form of spell checkers, snap check deposits, and simple Internet searches. These machine learning algorithms learn to be very good at specific tasks, and that’s why we call them Artificial Narrow Intelligence (ANI).

Recently, there has been a breakthrough in the form on neural networks. The idea is to connect a bunch of mathematical operations which we call neurons. On one side of the network are inputs, the pixels of an image, and on the other is a classification, the determination of whether the image contains a cat or a dog. Using very large datasets, we can make the network learn how to describe an image. The excitement around using neural networks has sparked an AI race between large tech companies, universities, and governments. The promise is a move towards artificial general intelligence with the potential to drive cars, program computers, control robots, and dictate financial markets.

By definition, the further we move towards Artificial General Intelligence, the more humans are going to be pushed out of jobs in favor of computers which are cheaper and perform better. In the next five years, it is likely that both the trucking and taxi industries will be uprooted, taking human drivers off the streets, putting traditional car makers out of business. Manufacturing will be dominated by robots. Even knowledge work isn’t safe.

For the people building machine learning algorithms, the focus has been on creating systems that are better than humans, systems that could potentially takes away jobs. This is a threat. Although AI technology has the capacity to improve everyone’s lives, initial indications show that it will destroy the lives of many. In their current form, these technologies are not intended make workers better at their jobs but to take the jobs away from the workers entirely. Taking away jobs isn’t a popular idea, and there needs to be a bridge to help ease the AI transition.

The simplest solution that has been suggested to help pad mass unemployment is a form of universal basic income. The work that is being done by machines will still be producing value, but it is how the value delivered by that work is distributed that matters. Universal basic income promises that every day people will still benefit from massive gains in AI technology. Universal basic income might work on a small scale, but to implement it on a significant scale quickly would likely create significant amounts of unrest. Giving someone money for not having a paying job does not compensate for the sediment that a robot is taking that job away. The technologies that we build now need to make work effortless, not take work away. It is much easier to accept help that makes life more fun than it is to accept losing your job. The AI transition will likely take longer than some people believe, but it will happen in our lifetimes and we need to be ready for the implications on our culture and society. Like every other disruptive technology before it AI will have good, bad, and unexpected effects on our lives, and it is important to think about the effect on culture beforehand.