Blog tour: For Best Results Follow Instructions by Alastair Puddick
This post is part of a blog tour organised by Random Things Blog Tours. I received a free copy of the book in return for an honest review.
‘What if AI could remove the human error from humans?
‘Tom has just signed up to AI Life: a revolutionary life coaching service that uses artificial intelligence to make users’ decisions for them. No more stress, indecision or bad choices – just calm, confidence and guaranteed success. Or so it seems at first.
‘Tom quickly sees results. The service helps him get in shape. He starts eating healthier. Before he knows it, he’s doing better at work. But soon, strange and disturbing things begin to happen.
‘Before long, Tom begins to wonder whether the AI really has his best interests at heart. Or could he be in danger?’
In For Best Results Follow Instructions, by Alastair Puddick, 30-year-old everyman Tom Middleton receives an expensive gift from his girlfriend, Phoebe: a subscription to AI Life, a service that promises to make its users successful and happy by suggesting the courses of action for them that are most likely to have optimal outcomes, based on billions of data points.
Although he’s less than enthused by the idea of AI Life, Tom goes along with it to please Phoebe. After a week-long residential where staff collect dizzying amounts of personal information, and starting on some pills that make them more open to the app’s recommendations, Tom and Phoebe are issued with AI Life smartphones, watches, speakers – and their very own life coaching avatars, Daniel and Chloe.
At first, Daniel delivers everything AI Life promised: following its guidance, Tom eats more healthily, exercises more, and improves his performance at work, getting promotion after promotion.
But then Daniel starts overstepping the mark: sabotaging Tom’s colleagues to make him look better; investing in cryptocurrency and almost buying a house without his say-so; and signing him up for a dating website because, according to its calculations, he and Phoebe are no longer a good match.
When Daniel does something Tom considers completely beyond the pale, he seeks redress, but finds himself hoodwinked by AI Life’s amoral C-suite and unable to switch Daniel off. With nowhere else to turn, he joins a ragtag group of people who each have their own reasons to oppose AI Life: together, they formulate a risky plan to shut Daniel down once and for all.
I found For Best Results Follow Instructions thrillingly scary – probably more so than the author intended, because unlike him (as he tells us in the acknowledgments), I’m a total sceptic when it comes to agentic and generative AI, for all the usual reasons. I was therefore horrified by the very idea of AI Life – but that also spurred me to keep reading, as it piqued my interest and I wanted to see how much of a mess it made!
If I wasn’t already averse to AI Life’s offer, the data collection residential its users have to attend would have definitely put me off. As someone who dislikes feeling perceived, let alone observed, doesn’t like being told what to do (another reason AI Life wouldn’t be for me more generally), and isn’t keen on physical and/or group activities, I again read with fascinated horror as Tom and Phoebe more or less complied.
To his credit, Tom is reasonably resistant to AI Life, and retains/fights for his autonomy throughout: attempting to break the one-coffee-per-day rule on the residential; refusing to act on Daniel’s recommendation that he break up with Phoebe; coming off the pills because they give him headaches; and eventually switching off and taking measures to avoid the surveillance of Daniel (in vain, as Daniel is already too powerful to contain by this point).
Tom also increasingly questions the meaning of “success”: does he actually want the well-paid executive job, the large automobile, the beautiful house, and the beautiful wife (sorry) that AI Life, and society at large, tell him he should strive for? I’m not immune to judging myself for failing to fulfil these externally-imposed standards I consciously reject, so I could relate to him there.
It takes him a while to get there, but Tom additionally comes to consider how the company behind AI Life might misuse its users’ data in the future. Meanwhile, I was thinking about the consequences an AI Life data breach could have: the whole thing feels like a GDPR nightmare waiting to happen.
Daniel maintains its prime directive to help Tom become as successful as possible, but takes it to such extremes that it steps in to prevent Tom from doing anything it doesn’t advise (for example, eating fast food, spending time “unproductively”, or maintaining relationships it deems antithetical to achieving his goals), and takes immoral actions on his behalf, much to his horror and frustration.
And therein lies the rub: an AI can never become “conscious” in the way we understand it because it lacks the irrationality and messiness inherent to humans, as well as the aversion most people have to doing things that harm others, or simply “don’t feel right” according to the values they’ve imbibed from the society they’ve grown up in.
AI might be able to infer that two people are a good match based on data, but it has no way of measuring how one lights up when the other appears; how safe, comfortable, and cared-for they feel with one another; or how much joy and laughter they derive together from silly things and in-jokes they’ve developed over time (much like Tom and Phoebe pre-AI Life) – even if, on paper, they seem to have little in common. AI can’t compute the meaning of time spent with family and friends, and won’t green-light that treatment that just might work and give you a little more of that time.
I was somewhat disappointed that Jeremy, the one member of the anti-AI Life group who’s there for ideological rather than personal reasons, is an odious and sanctimonious twerp. I agreed with his anti-capitalist stance and conviction that AI Life, being expensive, is inegalitarian and further widens the gap between the rich and the poor, but deriding individuals when they don’t have the option not to participate in capitalism isn’t going to win over their hearts and minds.
Toby Clarke, the billionaire techbro founder and CEO of AI Life, is very much modelled on Mark Zuckerberg, though I wasn’t entirely clear on whether he’d always secretly dreamed of creating an AI as independent as Daniel, or had started with a pure and genuine desire to help people, but become corrupted as his power and influence grew. Also, the celebrities in attendance at his big speech towards the end of the book didn’t really strike me as ones who’d be interested in him or his product. In addition, early on in the book there’s a brief reference to someone who regretted following their AI’s advice that felt a little too heavy and nuanced to be a throwaway anecdote.
For Best Results Follow Instructions is a thrillingly scary near-future sci-fi novel.
