Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
In his 1928 short story “The Machine Stops”, EM Forster imagines a ruined Earth where humans occupy isolated pods lined with buttons to serve their every need, from cold baths to literature, connected by an all-seeing communications machine. Even before the pandemic, this little-known story was traded among technologists as a prescient vision of the contemporary internet. A quote from it opens Shannon Vallor’s clarifying new book, The AI Mirror.
Vallor is a philosopher of technology who has spent most of her career in Silicon Valley’s Santa Clara University, latterly moving to the University of Edinburgh. Her first book, Technology and the Virtues (2016), was bathed in California sunshine. It gently asserted the relevance of “virtue ethics” — philosophical approaches dating back to Aristotle, Confucius and Buddha that centre around human qualities such as courage, moral imagination, honesty and empathy — in learning how to thrive in today’s technological age.
The AI Mirror extends this argument, with an urgent new call to, per the book’s subtitle, “reclaim our humanity in an age of machine thinking”. No prizes for guessing what’s changed: the release of OpenAI’s human-fluent chatbot, ChatGPT. Vallor’s new book sits among dozens published this year grappling with a world where it appears Forster’s literature button has become a reality. The AI Mirror stands out for its witty, crystal-clear exposition of the real threat from AI.
The author urges us to recognise that, far from being the handmaidens of a superhuman intelligence, ChatGPT and similar technologies underpinned by so-called Large Language Models (LLMs) are merely “giant mirrors made of code, built to consume our words, our decisions, our art . . . then reflect them back to us”. However real they seem, mistaking them for the beginnings of an Artificial General Intelligence — AGI, ie, machine sentience — is misguided, because “these mirrors know no more of the lived experience of thinking and feeling than our bedroom mirrors know our inner aches and pains”.
AI will put our future in jeopardy, argues Vallor, but not in the manner the “western media and political obsession” with AGI has imagined since ChatGPT’s launch in November 2022. Those who worry about sentient machines subjugating humans and imposing an alien and hostile value system on future generations are missing the point: AI is made in our image, and reflects the dominant value system of today. What we should be concerned about is how AI blocks our capacity to reinvent our values, “preventing us from knowing how to make a future at all”.
The values of our wealthy post-industrial society have brought us “to our current heights of scientific ingenuity”, but also to “the brink of planetary devastation”. The more power we cede to machines that can do nothing but reflect back these values, the less power we have to apply our own human capabilities and practical wisdom (what Aristotle called phronesis) to the problems that face us.
Vallor acknowledges her mirror metaphor is close to that of the mimicking parrot — famously able to trick humans into believing they understand more than they do — employed in the now infamous 2021 academic paper about LLMs, “On the Dangers of Stochastic Parrots”, which sparked a controversy that cost two of its authors their jobs on the ethics team at Google. It also marked the beginning of a shift in industry and government, away from incorporating ethics — and ethicists — in their strategies for allaying public fear about the dangers of AI. Vallor herself was only just out of a two-year visiting research position at Google at the time.
Public policy is now focused on AI safety, and some insiders talk about the “X-Risk” — existential risk — of AGI as being as urgent a threat as climate change. This, writes Vallor, is a “looking-glass fantasy” all too easily placed in the service of those seeking to defer action on the “real and imminent existential threats” of a warming planet.
The key to “human and planetary flourishing”, according to Vallor, is to recognise that we, humans, are the ones with the creative capacity to redirect technology’s course. Much as in Shoshana Zuboff’s groundbreaking 2018 book The Age of Surveillance Capitalism, the solutions Vallor proposes seem weak in comparison with the threats she so eloquently expounds. As with Zuboff, it will be left to others to work out how to take up Vallor’s cause.
The urgency with which Vallor states her case is not matched by any radical programme of action or bold challenge to entrenched political and economic power. But perhaps that’s the point: with their talk of sentient machines and human extinction, it is the X-Riskers that lay themselves bare to accusations of hysteria. Our job is to counter with a measured, and human, response.
The AI Mirror: How to Reclaim Our Humanity in an Age of Machine Thinking by Shannon Vallor OUP £22.99, 272 pages
Join our online book group on Facebook at FT Books Café and subscribe to our podcast Life & Art wherever you listen
Read the full article here