Whither AI?

Futurist Martin Ford – author of Rise of the Robots: Technology and the Threat of a Jobless Future – argues that artificial intelligence (AI) will shape almost every aspect of human life. Ford’s thesis is that advances in AI result partly from research in deep learning based on deep neural networks that imitate the structure of the brain. He interviews AI pioneers and thought leaders about AI’s future. His results will engage computer scientists, investors, and anyone interested in the history and future of artificial intelligence.

Today, AI informs the reality of everyday life and is on the cusp of affecting every aspect of the economy, society, culture and daily life. Ford envisions AI as becoming a general-interest technology that functions like a public utility.

The renowned figures who praise this book are a measure of Ford’s stature. Former vice president Al Gore called it “… an invaluable opportunity to learn from some of the most prominent thought leaders about the emerging fields of science that are shaping our future.” Eric Schmidt of Google said this is “…a unique and fascinating collection of perspectives from the top researchers and entrepreneurs who are driving progress in the field.”

Artificial Neural Networks 

Ford explains that most significant advances in AI in recent years – such as image recognition and language translation – result from the development and use of deep learning or deep neural networks. Neural networks are computer programs that mirror the structure and function of the human brain. 

Deep learning – at least so far – is the primary technology that has powered the AI revolution. Martin Ford

Ford credits the massive increase in computing power and available data for fueling neural networks that caused a revival of deep learning technology. He says simply that neural networks and deep learning are the heart of the AI revolution.

Progress

Google Cloud chief scientist Fei-Fei Li suggests that visual intelligence connects to virtually every function of human intelligence, including language, reasoning and decision-making. Fei-Fei cites the 2012 ImageNet competition for combining massive computing power with neural networks and launching computer object recognition.

According to Facebook vice president and chief AI scientist Yann LeCun, researchers must comprehend how human beings and other animals learn by observation. Until researchers come up with a model of self-supervised, unsupervised and predictive learning that explains how people acquire basic knowledge and a form of common sense, LeCun maintains, science can make little progress toward artificial general intelligence (AGI).

Integration

Judea Pearl, UCLA computer science professor, cautions that the statisticians who dominate AI assume data will reveal everything worth learning.

Neural networks and reinforcement learning will all be essential components when properly utilized in causal modeling. Judea Pearl

But, he insists, you cannot represent things you’ve never experienced (such as imagination and counterfactuals) with numbers or formulas.

Economic Impact

Ford discusses how AI will create jobs, but also is likely to automate daily household work, manufacturing jobs and tasks that professionals such as doctors, lawyers and stockbrokers perform. This may result in high unemployment and deepening economic inequality. Yet, Ford maintains, AI will generate more jobs than it will destroy.

James Manyika, chairman and director of the McKinsey Global Institute, agrees with Ford that AI will increase productivity and innovation.

Autonomous Weapons

Ford cites the risk of cyberattacks on AI systems and the immediate, dangerous threat of AI use for military purposes. He warns that AI-powered drones equipped with lethal weapons could inflict casualties without involving humans in the immediate decision-making process. He reports that, with this risk in mind, AI researchers and companies have issued a public pledge never to develop such weapons.

Ford lists another concern – that advanced AI-powered machines will evade human control and pursue goals in conflict with human aims.

One distinctive problem with very advanced AI systems…is that it presents not only the possibility of humans misusing the technology… but also the possibility that the technology could misuse itself, as it were. Oxford University philsopher Nick Bostrom

Google director of engineering Ray Kurzweil suggests that the problem isn’t that AI systems may develop goals at odds with human goals, but that human goals fail to align with one another.

Ford asked respondents to predict the earliest date by which AI has a 50% chance of being functional; the answers ranged from the years 2029 to 2200, with a mean date of 2099.

Omnibus

As in any omnibus of opinion, some of the voices here are revelatory, some are mired in cliché, and some are capable of only conventional wisdom. Yet even the last kind provides insight, because AI is still so new and its future so uncharted that it helps to learn which constructs and views of AI will shortly suffer obsolescence. So even the least perceptive interview subject – and all those gathered in these pages are articulate and expert – provides new thoughts for everyone save those deeply enmeshed in AI. Though Ford’s book offers a comprehensive overview for students, professors, philosophers of AI and futurists, it is not academic. The discussions are lively; no one addresses practical consequences without raising moral ones, and that makes this a rare, complete and valuable prism into the future.

If Ford’s approach speaks to you, you will enjoy his Rise of the Robots. Other books offering worthwhile related insights include Melanie Mitchell’s Artificial Intelligence, John Brockman’s Possible Minds, Stuart Russell’s Human Compatible and Gary Marcus and Ernest Davis’s Rebooting AI.