Mr. Russell makes a delicious excursion into the philosophy of utilitarianism and the psychology of preferences. How do we balance the wants and well-being of different people, or the same person across time and mindsets? Is there even a singular self with coherent desires? Mr. Russell’s exciting book goes deep, while sparkling with dry witticisms.
Russell has created the right guide at the right time for technology enthusiasts seeking to explore the primary concepts of what makes AI valuable while simultaneously examining the disconcerting aspects of AI misuse.
Russell sidesteps the biggest problem: consciousness. In this area, he says, 'we really do know nothing, so I’m going to say nothing', adding that 'no behaviour has consciousness as a prerequisite'. No behaviour, maybe, but if machines are to be purposeful, as well as merely mimic or obey human purposefulness, consciousness might be crucial ... This is not quite the popular book that AI urgently needs. Its technical parts are too difficult, its philosophical ones too easy. But it is fascinating, and significant.
What is certain is that Human Compatible marks a major stride in AI studies, not least in its emphasis on ethics. At the book’s heart, Russell incisively discusses the misuses of AI ... It might seem surprising, in a solid, cautionary account of contemporary misuses and abuses of AI, that Russell fails to do justice to current boots-on-the-ground benefits. These are already helping people to tackle challenges such as climate change, the biodiversity drain, disease detection and disaster relief ... Russell, however, fails to convince that we will ever see the arrival of a 'second intelligent species'. What he presents instead is a dizzyingly inconsistent account of 'intelligence' that will leave careful readers scratching their heads. His definition of AI reduces this quality to instrumental rationality ... Still, Russell all but admits that instrumental aptitude is not enough to account for the full gamut of intelligence capability. Sacrificing the coherence of his own definition, he hedges his bets ... merely rehearses more than 60 years of unanswered criticisms, intractable shortcomings and repeated failures ... Russell ultimately falls prey to the techno-solutionist idea that intelligence can be treated as an ‘engineering problem’, rather than a constraining dimension of the human condition that demands continuous, critical self-reflection ... Russell’s book does, however, sound an important alarm bell, although it is not quite the one he intended. The real control problem isn’t managing the coming of transcendent superintelligent creatures. More critically, it has to do with reining in the triumphalist creators who may be developing increasingly 'autonomous' AI technologies under the auspices of the misguided definition of intelligence that Russell himself champions.
... authoritative ... An accessible writer, Russell is addressing nonspecialist readers, and he largely succeeds, although some will find his detailed explications challenging. (Refreshingly, he appends the most technical text) ... A strong case for planning for the day when machines can outsmart us.
