Previous month:
December 2018
Next month:
March 2019

Two faces of Artificial Intelligence

At its core Artificial Intelligence is about teaching computers to do what humans can do with the expectation that computers will do those things better. Examining the history of technological progress it is possible you will not live long enough to see Artificial Intelligence change the world. Assuming AI is something that will change the world and that is in no way assured.

In a best case scenario Artificial General Intelligence (AGI) will be system capable of tapping the sum of human knowledge to answer questions which are currently beyond us and generate ideas which we are incapable of. AGI relies on a breakthrough yet to be made so in the near term we can expect that slivers of task specific Artificial Intelligence will be embedded into software, services and products in the same fashion that databases are now embedded in the such things. There was a time when the very idea of a database in your home or in your hand was ridiculous but now task specific databases are embedded throughout a multitude of devices people own or interact with.

In 2014 one of the more popular books bought by technology industry executives was The Second Machine Age. In The Second Machine Age the authors, Brynjolfsson & McAfee, propose that advances in software design and computational power are doing for thinking what the steam engine did for manual labour. To the authors true innovation is in combining things that already exist in different ways to create new outcomes. Add AI to a vehicle and gain the benefits of self driving vehicles.

This is an optimistic book which takes great pains to point out that quality-of-life improvements will not be distributed evenly and there will be losers from this ongoing cognitive revolution. In the opinion of the authors the greatest gains in productivity, wealth creation, and social good lie ahead but we must remember to bring everyone along. This is a book where technology not only saves the world but makes a better for everyone and regardless of the strength of the ideas proposed between its covers that is a very appealing vision for people working in the IT industry.

The antithesis of the second machine age would be The Rise and Fall of American Growth. Written by Robert J. Gordon this proposes that life began improving dramatically for people through a series of great inventions, such as electricity and the networked home. It is an example of our focus on Information and Communication Technology that the idea of a networked anything would involve Ethernet but in this case the networks are those of electricity and indoor plumbing. Electrification brought light and the mechanical automation of repetitive chores into the home, while indoor plumbing provided freshwater for consumption and as importantly increased public health through better sanitation.

In Gordon's view the century of unprecedented growth between 1870 and 1970 was an outlier and not something that will be easily repeated. Using the example of the internal combustion engine Gordon proposes that important inventions do not have an immediate impact and must be adapted and disseminated. In the case of the internal combustion engine it took nearly 50 years before tractors replaced horses on farms. The greatest inventions have shown that the process of dissemination is slow but provides steady increases in living standards over a long time.

In Gordon's research the development of the Internet and other related Information and Communication Technologies created a surge in productivity between 1994 and 2004 which then tailed off dramatically. Unlike the inventions in the century of unprecedented growth the dissemination of the Internet did not create a significant increase in living standards. In Gordon's view the algorithm is no match for the assembly line when it comes to making people's lives better. Artificial intelligence may be able to quickly identify what is a cancerous growth in a patient and what is not, but delivering untainted water to where billions of people live and taking away their waste has saved and will continue to save orders of magnitude more people.

This is not to say there is no value in Artificial Intelligence but Gordon's view is that we have already exited an unprecedented cycle of intellectual achievement and quality of living increases throughout the 20th century, and we are now returning to incremental increases in living standards we have known throughout human history. Artificial General Intelligence, where computers can truly do things humans can do and do them better, still eludes us. Accepting that any breakthrough in this field would take time to be adapted and disseminated we see there are decades between the creation of the first AGI system and the adoption of AGI systems throughout society.

Artificial Intelligence may change everything and introduce the long boom of the second machine age, or with a lot of the hard work to increase living standards already done in the 20th century it may just provide incremental improvements to our lives by being task specific. But the clock does not start ticking on the societal impact of artificial intelligence until we have a major breakthrough.

Today AI can beat the best DOTA2 players, OpenAI Five playing 180 years worth of DOTA2 games every day and using what it has learned to demolish human players in the arena, but if you change the game those simulated decades of experience become worthless. The breakthrough we are looking for may come from gameplaying artificial intelligence but that breakthrough is not artificial intelligence which can only play games. The clock hasn't started yet and the decades required for adaptation and dissemination will not begin until it does.

How will we know when artificial intelligence has made a true impact on society? When it starts telling us things we do not like to hear.