What If We Got AI Right? By Eleanor Drage review – avoid the apocalypse | Books on science and nature

TAI sociologist Eleanor Drage believes that for artificial intelligence to thrive, people need to recognize the “personality” of AI. He means that although we often talk about AI as if it were a mysterious, intangible thing, it comes from long hours of work, in silicon and quartz mines, microchip factories and data centers. If citizens want to wrest power from the tech companies that are building AI, it helps to better understand what they are selling us. This also means seeing through the marketing and hype: “the cloud”, the invisible, remote place where your photos are stored, is someone else’s computer; AI “hallucination” is a system error or data entry error; Intelligence is a broad term – and one that can be misleading.

The story in Silicon Valley is whether AI will save us, usher in a new era of development and human longevity, or it will destroy us all. Drage argues, persuasively, that big tech cannot live up to its promises as it continues to chase profits and power above all else, while also serving the interests of the few. Meanwhile, the cultural obsession with the apocalypse created by AI also diverts attention from the discussions we can have about how to make AI safer – by ensuring that citizens have more control over how their knowledge is used, and how AI models are trained and controlled, for example. The apocalyptic vision of Tech titans also creates the false impression that AI security is more important than climate change, global conflict and inequality, three problems that are already causing great suffering and that, if addressed, would reduce many of the risks associated with AI.

Drage, an associate professor at the University of Cambridge’s Leverhulme Center for the Future of Intelligence, has conducted a study that challenges recruitment strategies using the power of AI, and reveals the dangers of incorporating AI into law enforcement. Many of his criticisms of big tech will be familiar: that AI algorithms perpetuate and enforce pre-existing prejudices; that for all their empty words, promising to be “transparent” and “accountable”, the tech giants have become too big to care about ethics; that a few directors have gained too much power; and that the release of AI is being done without our permission and without considering the health of the planet. But his writings are very intelligent, and his analysis is very intelligent about, for example, how we define intelligence and why this is so important.

However, at times, Drage’s arguments are quick and sloppy. “It’s just an illusion that the workforce is disappearing: work is just changing, from artists to data writers and from salespeople to engineers,” he once wrote. But it is not clear that artists and “businessmen” will be rare in the coming AI era. It is not as if data writers working on poverty in Kenya would have been artists 10 years ago. Likewise, given the importance he places on raising the diversity of AI, it’s disappointing that he dismisses unconscious bias as “just another way of looking at the offender rather than the victim”, as if the goal was ineffective. He writes that people are ready to have “compassionate” discussions about equality, adding that these discussions also require precision and clarity.

The title of the book suggests that Drage wants to offer another vision of what AI could be. He gives several examples of the kinds of AI projects he thinks are set to replace high-tech, and many of them are surprisingly complex. They include an app called Mumkin designed to facilitate critical discussions about female genital mutilation (FGM) in India, and Kuini, an AI interface designed to help Māori women quit smoking. In both cases I wondered what value AI brings to these efforts – especially given all that Drage has discussed about the impact on the environment and the supply chain. Is the program more effective than training community members to become mentors, or support groups and co-workers?

Drage describes the tools he created to help AI companies succeed and comply with EU regulations as a “cornerstone of utopianism”, because they change the way we look at how AI is designed and built. In his opinion, utopia is not a destination but “a way of working together to create a better world”. It may not be fair to expect a real vision, but I would be interested to know how AI, in particular, can help us build a better future. He did not mention things that many critics of AI welcome, such as its potential to improve drug discovery and disease diagnosis. What if we got AI right, the book asks. After reading it, I can’t say I really know.