A Brief History of Artificial Intelligence: What It Is, Where We Are, and Where We Are Going by Michael Wooldridge. Michael is Head of Department of Computer Science and Professor of Computer Science at the University of Oxford.
Is artificial intelligence intimidating to you? Above all this is a very easy, enjoyable book. So, Michael states in his introduction “I’m writing a popular science introduction to artificial intelligence.”
Accordingly, Michael has researched artificial intelligence for over 30 years. He is focusing on multi-agent systems drawing upon ideas from game theory, logic, computational complexity, and agent-based modeling.
A short history begins with Alan Turing’s work in 1935 at Cambridge during World War II. This is beyond America’s cultural understanding of Turing’s life from the 2014 movie The Imitation Game. Alan Turing actually defined artificial intelligence.
Machine Learning
Chapter 5: Deep Breakthroughs, addresses why Google acquired DeepMind Technologies, a British-based research laboratory in 2014. Founded in September 2010, DeepMind was introducing a term bounced around a lot: Machine Learning.
There is certainly a great misunderstanding regarding machine learning and deep learning. Additionally, Micheal’s efforts are to be complimented in making this topic understandable.
This may be one of the best chapters as readers learn about a subset called neural networks. So the reason Google acquired DeepMind their ability to solve the puzzle of intelligence. The key was how DeepMind was able to apply machine learning to the popular 1970 computer game Breakout:
After a few hundred more training rounds, the program becomes expert: it doesn’t ever miss the ball. And then something remarkable happened. The program learned that the most efficient way to accumulate points was to “drill” a hole down the side of the wall so that the ball was above the wall: the ball would then quickly bounce between the wall and the top barrier, without the player having to do anything.
This behavior was not anticipated by DeepMind’s engineers: it was autonomously learned by the program.
p. 171
Then DeepMind followed up by building a more famous artificial intelligence system called AlphaGo which played the ancient Chinese board game Go.
Artificial Intelligence in Healthcare
Michael addresses the short term future of artificial intelligence by DeepMind’s AlphaFold system that is addressing protein folding used to treat Alzheimer’s disease.
Therefore, use of artificial intelligence in healthcare has new possibilities. Similarly, at the University of Oxford new research is revealing signs to identify the onset of dementia simply by how we interact with our smartphones. This week BBC posted DeepMind uses AI to tackle neglected deadly diseases.
So what could go wrong?
Above all, Michael is fully aware of how artificial intelligence technologies can go wrong. There important issues are keenly outlined in closing chapters:
Chapter 7 How We Imagine Things Might Go Wrong
Chapter 8 How Things Might Actually Go Wrong
Unquestionably, there is no doubt Michael fully understands how artificial intelligence can manipulate. This is especially refreshing to understand first hand his understanding of technology pitfalls. At the same time, Cathy O’Neil’s excellent book Weapons of Math Destruction is addressing errors with big data.
In conclusion I certainly found this book fascinating and insightful. Understanding this often mentioned topic as Michael reveals, helps bring a sense of understanding to applications of artificial intelligence to driverless cars and other markets and organizations.
J.P. Morgan | The Evolution of AI
NOUS | Michael Wooldridge on the History and Future of AI
Philip Davies | The Road to Conscious Machines
The Alan Turing Institute | History of MAS research in UK