Thinking Machines: The Inside Story of Artificial Intelligence and Our Race to Build the Future
by Luke Dormehl
(W H Allen. £14.99)
Luke Dormehl’s accessible and congenial book considers the development of intelligent machines, and the impact they are having on our lives.
Almost 70 years after British mathematician Alan Turing wrote the first programme that would enable a machine to play chess, machines can beat grandmasters; they are capable of learning without preloaded programmes; they compose music, invent recipes, and drive cars; a machine wrote an obituary of AI (Artificial Intelligence) pioneer Marvin Minsky after his death earlier this year.
Advances in AI reflect progress in neuroscience. The more we know about the human brain, the more intelligent the machines we create.
Once scientists fully understand how the brain works – that day is still far off – they may be able to create machines that simulate its workings. The final step would be to ‘upload’ human minds into machines, freeing them from the physical body, and effectively making them immortal.
Dormehl is fascinating on what we might call ‘artificial creativity’. Given enough information machines learn to identify patterns in music or images, and use these patterns in turn to produce new music or images. Or put simply, they come up with new ways to combine old ideas.
The short machine-generated account of a bank heist that Dormehl reproduces may not rival War and Peace but it nonetheless works as a story.
Academic
He has interviewed an academic who has uploaded all of the Beatles’ music into a machine, in no particular order. After identifying changes in the band’s musical style, and without recourse to the internet, the machine listed the releases in correct chronological order; it can also predict what a Beatles’ album from 2026, might sound like.
As also happened during the Industrial Revolution the rise of the intelligent machine has created new forms of inequality and exploitation. Google, for instance, uses the work of Hanna Luetzen, who has translated the Harry Potter books from English into Danish, to improve its English-Danish machine-translation service, without paying her anything. The status of data is a pressing issue. If data is the oil of the digital economy, says Dormehl, then “we need to put a proper valuation on it”.
Machines are intelligent, but in a limited way. They are not self-aware. They will remain our workhorses until the day scientists fully understand and can replicate the functioning of the human brain. The development of conscious, thoughtful rather than thinking, machines, raises profound questions.
Can a machine learn, or be taught, to behave according to the moral and ethical codes that largely restrain humans from harmful actions? What would be the status of sentient machines? Subservience to, or equality with, humans? Could a machine refuse to obey a command given to it by a human?