Another AI reading list: these books are worth your time

A couple of years ago, I published an AI reading list with five books I found helpful. As lots of great books came out recently, this post has five more.

I realise this post comes just after most return from their summer holidays, but also, that's why I had extra reading time. It checks out.

The common thread in these books is that they provide a critical look at the current wave of AI: the companies, the marketing strategies, the funding and the dangers. They aren't “negative”, that'd be unfair to say—they carefully assess what's going on. I am glad that there's plenty of well-researched analysis available, especially in a week in which we saw another series of billion-dollar investments in AI and data centres announced.

One book didn't make it to the list, as I haven't gotten around to reading it, but based on the author's blog posts I look forward to digging in: The Intelligence Illusion by Baldur Bjarnason. Its print edition has just been released too.

The AI Con

That modern LLM's are really “synthetic text extrusion machines” is the boring truth that both “doomers” and “boomers” seem to forget: the people who make a lot of noise about either AI could destroy the world or about that is the best thing since sliced bread, respectively. This discrepancy between how AI is presented by those who (want to) benefit from its adoption, and what the technology really accounts to, even if impressive, shows that it is overhyped. While causing a large number of harms. And setting out to cause some more. This and more is covered in The AI Con, a book I looked forward to ever since its announcement, partly because it was co-written by professor Emily Bender, coiner of the phrase “stochastic parrot”. Some reviews called it condescending and overly reliant on ridicule. There is some of that, and I do miss the bit where the valid use cases of text extrusion are acknowledged. But it doesn't make The AI Con less of a must-read for anyone involved in making decisions about AI.

Empire of AI

To understand the current rush towards AI adoption, we must understand the companies and how they operate. Empire of AI by Karen Hao, zooms in on OpenAI. On its founder, Sam Altman, who was used to fly private jets before starting this particular company and was mentored by Peter Thiel, the influential and controversial billionaire.

Hao explains OpenAI are very keen on making ‘artificial general intelligence’ and raise billions to try and solve climate change and healthcare with it, while they don't have that actually would mean… “At this point”, she says, “AGI is largely rhetorical—a fantastical, all-purpose excuse for OpenAI to continue pushing for ever more wealth and power”. Hao notes that intelligence itself isn't even well defined and ‘artifical intelligence’ was even in 1955 a marketing term coined by John McCarthy who realised his original phrase ‘automata studies’ didn't fly.

Empire of AI shows how OpenAI positions itself and thinks of itself, as a place that is somehow destined to develop AI “safely”. And how they change: idealistic, academic, open and research-y first, billion dollar business with colonialist tendencies later. It's an Empire, Hao explains: it excites talent with a grand ambition, grabs capital to remove what they see as roadblocks and has a mission so vague that it can be constantly reinterpreted.

Unmasking AI

A lot of the books in this list talk about the “collatoral damage” of producing and deploying AI systems. One thing that is clear about this damage is how unevenly it is distributed. The facts are disturbing, for instance with AI being deployed in war zones, and in algorithmic systems that affects people's lives directly. In Unmasking AI, Joy Buolamwini, founder of the Algorithmic Justice League, goes into what's happening and how it's intersectional. For discrimination in algorithms, she coins the phrase “coded gaze”: “the ways in which the priorities, preferences, and prejudices of those who have the power to shape technology can propagate harm, such as discrimination and erasure”. Buolamwini takes the reader along with her personal quest to do something about this phenomenon, which at times becomes a bit memoir-y, I liked that.

Resisting AI - an anti-fascist approach to artificial intelligence

Resisting AI by Dan McQuillan, who also writes great blog posts, absolutely resonated. Artificial intelligence is a gift to the far right, the book explains. For various reasons, including because it is great at drawing borders, a practice that benefits those in power much more than people who are marginalised by them. Therefore, we need to find ways to resist it, with worker groups, and by imagining alternatives that aren't just plasters on today's technology. It was published a few years ago, and since then we may have missed the chance to properly resist. It seems less impossible now that AI has become omnipresent, now that even simple internet searches trigger generated “summaries” by default and meeting software continuously begs to insert its bots. But imagining alternatives is still a way to resist, and people are putting that into practice.

The Tech Coup

Some politicians aren't techy enough to get the mechanics of what tech companies are doing to the world, which shows in how they try (or fail) to regulate. Not Marrietje Schaakje, a Dutch former Member of the European Parliament and now researcher at Stanford University. In the Tech Coup, she shares lessons from 15 years in tech policy. This book explains how technology threatens democracy and suggests how to mitigate, going beyond the “tech is bad” other books offer, with a clear “here's how we can make it less bad”. The book is not on AI specifically (it's discussed), but covers some of the companies that are selling it and how they go about that. How they are great at framing things, pretend to self regulate, get entangled in actual wars, and, in the case of Microsoft, open offices inside the United Nations to get a front seat in shaping technology recommendations.

As a bonus, similarly not about AI but about policy: I also loved Careless People by Sarah-Wynn Williams, it describes similar mechanisms, but from the inside of industry: the headquarters of (then still) Facebook. Great, shocking and painful anecdotes from someone who worked close to Mark Zuckerberg and Sheryl Sandberg, if you needed another reason to distrust Big Tech's PR strategies.