By Matthew Ambrose FIEP | Employability Consultant & Project DEEP UK Lead | Ambrose Business Services Ltd
In The AI Con, Emily M. Bender and Alex Hanna deliver a sharp, uncompromising critique of the artificial intelligence industry.
Their careful debunking of myths surrounding the various technologies packaged as ‘AI’, and exposure of the venture capital-fuelled marketing hype surrounding its growth, provides a punchy and accessible challenge to what already feels like an inexorable slide into an AI dominated world.
Bender, a linguistics professor known for her incisive critiques of language models, and Hanna, a sociologist and former Google researcher, bring both technical expertise and lived experience to bear in the construction of their central argument: the term “artificial intelligence” is misleading, and the hype surrounding it serves corporate interests at the expense of the public good, with its deployment often reinforcing or exacerbating existing social inequalities.
And while there is plenty of evidence to back these sentiments, the relentlessness with which they are pursued in the book renders it somewhat imbalanced and lacking in nuance, with personal politics sometimes clouding complex subjects (such as the dismissal of any debate about population trends as “eugenicist”).
A central argument in The AI Con is that the public are being fed misinformation about AI on two fronts. Firstly, that AI really does represent machine sentience, and secondly, that we must guard against an impending AI doomsday, with the robots taking over and dispensing with humanity. At the extremes, these narratives do pop up from time to time. But it must also be pointed out that no-one with even a cursory understanding of AI technologies really believes either of these positions to be true. In fact, I was recently at a large AI conference in London (full of industry leaders, AI experts and end point consumers) at which neither subject was even remotely raised. This and other similar examples leave the reader feeling like the book is, at times, chasing ghosts.
The truth of AI (a collection of statistical tools built on massive datasets) is more mundane, but that does not mean that we should not heed many of the warnings contained within the book. The data used to train AI is often scraped without the consent of the copyright holder. Algorithms are bound to discriminate if, for example, they fail to accurately translate a regional dialect.
And ‘automating jobs away’ can do real harm if there are no alternatives for the workers it displaces. But the Bender and Hanna’s tendency to place all the responsibility for these potential harms at the door of the AI companies themselves is also misplaced. The end user has choice and agency in how these tools are exploited, and it is clearly possible, within the right framework and with well-constructed principles in place, to utilise AI technologies right now in ways which support human flourishing, rather than inhibiting it.
In summary, while readers would be wise to heed the central call to action (resisting passive acceptance of AI hype and instead demanding systems that serve people – not profit), it is all of us who will shape the future of AI and how it is deployed.
The AI Con is indispensable for those who wish to better understand the technology, its potential uses, and the pitfalls we must all avoid to ensure a fairer, more productive future for all. The future isn’t written in code – it’s ours to shape.
ABOUT THE AUTHOR
MATTHEW AMBROSE FIEP | Employability Consultant & Project DEEP UK Lead | Ambrose Business Services Ltd
Matthew Ambrose is a specialist consultant with 19 years of experience in employability and justice. Following a career with Work Directions / Ingeus and the Reducing Reoffending Partnership operating across multiple roles and contracts, Matthew has spent over 7 years as a contractor supporting providers to mobilise contracts and improve performance and technology.
He is also the UK lead for Project DEEP, the first of a growing global partnership between the IEP and providers, with the primary aim of running a series of ‘test and learn’ projects and sharing the results across the industry to benefit practitioners and participants.