Artificial intelligence is basically the ability of machines to learn, react to things and generally behave just as good as humans. The biggest fears is that with advancements in AI, machines will become more like rational agents with ability to perceive their environment even better than humans. Two of the leading entrepreneurs that are on two sides of this argument. I have been following keenly. On one hand is Mark Zuckerberg who says that AI is a good thing and people should not fear it. Elon Musk on the other hand is one of main people who argues that AI should be closely monitored, if not it could lead to human extinction as we know it.
In this post I summarize their arguments.
In an interview with president of Y Combinator, Sam Altman, Mark explains why he is passionate about AI. The interview is here. He says that AI is already with us especially with the onset of self-driving cars. In addition, with more advanced machine learning the doctors will continually have more power to easily and quickly detect diseases rather than the current process of continuous tests. This is a good thing. His basic argument is that AI could end up helping even more people and therefore instead of being scared by it, it should be embraced.
In another interview with Axel Springer Zuckerberg explains that history shows that people always tend to be worried about a new technology and it has turned out okay afterall. In the above interview, Mark says that there are two types of learning: supervised and unsupervised learning. Supervised learning is that we all have grown up by. For example, telling a kid this is cat and the kid knows this is cat. It’s about pattern recognition. But now with AI, there is unsupervised learning. That is learning through refining something in your head and making a model and working it out and expect it to happen in the world based on your action. AI will help us do that. Therefore instead of building machines that eventually harm us, Mark thinks that unless the software engineers mess something really big, we should be able to control the machines. In any case, machines are built in accordance to human knowledge. Even today there are machines that can do more than human but they were built with that intention. So this is the same case with AI. Similarly, he says that just because we can build machines that can do some things better than humans does not necessarily mean that they can learn new domains and connect different types of information. In essence machines can still be controlled. He gives a good example of someone building the airplane in 1800s. If they thought about safety of planes crashing before even building the actual planes then they could not have been able to invent the aeroplane in the first place. This is same case with AI. Build fist and then think of safety measures, not the other way round.
The points raised by Mark are the exact reasons why Elon says we should be careful with AI. In this clip appearing on fortune magazine , Elon says that the problem with AI is not that it will not follow the will of the people who built it; it will follow the will but if it not well thought out it could have a catastrophic outcome. This has led Musk to be on the forefront in founding a nonprofit, OpenAi that ‘aims to democratise AI’ so that it is not used for wrong reasons. Musk says that AI will create machines so powerful that humans will find it hard to keep up. He even goes ahead to suggest that machines will be god-like creatures and humans will need to implant ‘neural laces’ in order to keep up. Science fiction stuff!.
All in all, we are already living in a world with AI in simpler forms that are set to develop with time. Ai is being used to comb personal information you feed on places such as search engines to predict patterns,marketing recommendations, financial information etc. This is based on a report by Bank of America on AI. In fact, Google is already using machine learning techniques to refine search engine results. Furthermore, Google Now and Cortana on Android and Windows are like virtual assistants that help in finding useful information. This is spreading to driverless cars-Google, Uber and Tesla (autopilot). Retailers such as Amazon could also be able to detect what you need before you need it based on your past shopping patterns.