AI: Is it Artificial or Intelligent?
October 5, 2018 in
There are lots of scare stories around at the moment about demonic robots stealing our jobs and wiping out the human race. More positively some people are predicting a future where we no longer need a job. Robots will do all the productive work and we get paid a minimum basic income instead of wages, leaving humanity free to follow its dreams or just waste more time buying junk or playing golf! Either of these dystopian or utopian versions of society may well be possible but both are unlikely. History teaches us that we generally muddle along for the most part avoiding annihilation. However, the development of AI isn’t just hype, and at some stage it will have a big impact on society (just don’t expect it to change overnight). Industrial revolutions take longer than that, even with Nano second computing power! So should we be scared, or should we embrace this change? While we can’t stop change, societies can direct and regulate innovation if they choose to. In most societies throughout history new technologies, such as the internal combustion engine, aircraft or radio and television broadcasting, have been controlled by national and international regulations. We are not helpless and we can shape the future.
There are many practical uses of AI, but perhaps we should first consider what it is not. This isn’t an easy task because there is no accepted definition of what “intelligence” is! Psychologists and other scientists can see much of the biological mechanics of thought but still lack an understanding or explanation. Human cognition seems to require a sense of perspective and consciousness, not just an ability to process or learn from information. If this is correct then we haven’t seen any evidence that AI is close to approaching real intelligence. What it is very good at is processing huge amounts of data and reacting to changes very quickly. That is why computers can beat the best humans at Chess. Complex games like this have what is, in a practical sense, an infinite number of possible moves but because they are all rule based a computer can work through them much faster than a chess grandmaster. Is this intelligence? I don’t think so, but it is very fast and very efficient processing. One day science may establish that human intelligence is just very fast processing, as the Turing principal suggests, but for now we just don’t know. What we do know is that AI uses rule based algorithms designed by humans. Machine learning, whereby AI reacts to data changes to adapt its outputs, is also based on algorithms. From this perspective AI isn’t revolutionary at all. It is based on algorithms, a mathematical technique used by Greek and Arab scholars over 2 millennia ago. What has changed is the speed of calculation and the ability to process huge amounts of data very quickly. So, while this is clearly the result of human intelligence, to my mind it isn’t intelligent in itself while it probably is “artificial”, i.e. made by the skill of some very clever people and not by magic or demonic robots!
So how can we apply this non scary, non-revolutionary but very useful technology? The potential to improve our living standards is huge, from autonomous vehicles to more efficient smart buildings and data centers AI is going to change the way we live and work. Just as the flint and the internal combustion engine did. The danger of AI is not that it will assume consciousness, or break Azimov’s robotic principle and do harm but that humans will direct it and use its power to do harm. This should not stop us from progressing with new technology. Unfortunately, unlike AI, humans are scary and we need all the help from technology we can get to gain a better understanding of our world and limit the damage we do.