We have so far been looking at evolution from the perspective of Biology and Geology, but the principles of evolution can help us understand other kinds of change as well. For example, evolution can tell us a lot about how computers and artificial intelligence are developing now, and where they may be going in the future.
Over the course of billions of years, evolution has created many forms of intelligent life. Migrating birds and butterflies can find their way over thousands of miles. Pets know exactly how to play their owners for food and attention. And humans have built sophisticated tools and machines that compensate where we lack in physical ability and allow us to live almost anywhere in the world and even land on the moon.
All of this happened incrementally, with each new species and generation changing how they could behave in their environment. But thanks to recent advances in technology we can now create entirely artificial, intelligent systems that some fear may even prove dangerous to us as a species in a few decades’ time. But how do we even know what intelligence is?
Imagine this: In 30 years you will be chosen to decide if we have truly developed a super-human artificial intelligence. Would an IQ test do? How would you decide if a work of art was created by a human or a robot? How would you know you are in a conversation with a human and not Siri? In many ways, deciding if an animal is intelligent is very similar to this problem. Would you agree that cats, dogs and dolphins are smart, but flies, fish and snails are not? Why? What behaviour do they show that is missing from today’s computers?
Let us try to narrow this down a little bit and start with abilities that are super-human but which few people would call intelligent. For example, a computer can calculate 321 x 134 in a matter of milliseconds. Your phone can give you directions to almost any point in the world. And artificial neural networks can create shockingly realistic images of landscapes and people that do not exist (see links below). The problems we can solve with these tools are arguably very complex and would take considerable amount of time and skill for humans to solve, but do not necessarily require intelligence. Similarly, the fact that birds can fly, fish can live in water and leopards can run extremely fast does not make us think they are more intelligent than us, even though they can do things that humans cannot.
It gets a little more complicated when we look at some more recent advances in AI: The London-based company DeepMind has built AI systems that can beat even the best human players at difficult strategy games such as Chess, Go or the computer game StarCraft (see link to video). This already comes a little closer to what we might require from an intelligent agent: to have the ability to act and react. Perhaps what we mean by intelligent behaviour is not any particularly high level of performance, but flexibility, goal-directedness, and the ability to create something new out of old parts. This is what researchers call Artificial General Intelligence, or AGI. Arguably, we are still quite far away from achieving this.
This endeavour involves scientists from almost any discipline, for example computer science, psychology, neuroscience, philosophy or economics. You may be surprised to see psychology and neuroscience in this mix. They play an important role because they tackle some of the questions discussed above head on and try to understand how particular biological systems can be intelligent. Their insights can then be used to improve artificial systems. In fact, the co-founder and CEO of DeepMind, Demis Hassabis, holds a doctorate in cognitive neuroscience. Similarly, tools from AI have helped us better understand the human brain and the sophisticated behaviours it can produce.
Much ground-breaking work remains to be done – would you consider studying one of the disciplines involved? Which one?
Explore the subject further…
“How to build a brain from scratch”: material of an Oxford Experimental Psychology undergraduate course by Prof Summerfield
What could be the dangers and benefits of building an artificial general intelligence? How might such a system react if you want to unplug it? Write a 300-word essay in which you consider these questions.
Dr Keno Juchems, Junior Research Fellow in Psychology
I have a background in psychology and cognitive science, but my current interests are better described as computational neuroscience. I am interested in how people make decisions about things they consider valuable (in terms of money or useful to complete a goal). I am currently investigating how people prioritize certain goals over others (e.g. when deciding which item on a to-do list to work on first) and how irrelevant goals (e.g. all other items on the to-do list) influence progress towards their chosen goal and their ability to plan. I use these questions with the aim to better understand the brain, particularly the prefrontal cortex.