USE IT OR LOSE IT

The Challenge of Artificial Intelligence

I used to carry about 100 phone numbers in my head: family, friends, my office. Now I carry just one: my husband’s. I can do this because I have effectively outsourced my memory to my phone. I’m comfortable doing so because I backup my phone and if, in a real emergency, I don’t have it, I can call my husband.

Other kinds of outsourcing I am not happy with. No matter how well AI agents might get to ‘know’ me, I wouldn’t want it to email my friends or fix a time I can have lunch with my daughter; I’d regard it as crass and impersonal—because it would be crass and impersonal. In the same way, podcasters who invite me onto their show, and then invite me to schedule myself don’t get a reply: I have zero appetite for so transactional an exchange.

The smarmy automated invitations to save me time by letting AI summarize long documents (wow at least 10 pages!) just make me mad at the addiction-peddling of Microsoft, a company that, if it truly respected my time, would devote real resources to improving their messy software. Suggesting that AI would derive from a long document exactly what I would get from it is flagrantly dishonest and implicitly insulting. AI does not know what will pique my interest or attention, what I’m looking for or what I need to know today. And how something is written tells me much that may be inchoate but nonetheless relevant to me.

I also want to write for myself. Writing is a form of thinking; it is as I’m shaping a sentence that I’m teasing out precisely what it is I believe, and what feelings and thoughts I seek to communicate. It is when I’m editing that I identify connections between ideas, people or attitudes. It is in re-reading that I notice, and may modify, a tone that isn’t quite right—thus making me consider carefully my relationship between reader and writer. Writing is how I develop and discover what I believe to be true.

So it saddens and alarms me when I hear rising executives and their colleagues enthusing about AI. “I love it””” one said to me. “I’m a rubbish writer and its emails are so much better than mine.” I understand their workload may force them to prioritize speed over effort, but how will they learn to express themselves, or to discover what really needs saying? How will they develop the discernment to distinguish between the formulaic, the emphatic, the personal or the inept?

At a conference last week, an executive in the audience asked what he should do when AI made a better decision than he did. But how would he know it was better? Judgment develops finesse by thinking through choices, debating alternatives, making a choice and watching what ensues. Outsourcing decisions sacrifices learning and insight. For me, the gold standard of good decision-making is a conclusion that even those who don’t agree with it can live with it, because they trust the process. But how will anyone learn to identify, let alone reach, a decision that meets that standard? How can you explain or defend choices you did not make?

Use it or lose it. Lots of AI advocates scoff at such reasoning, drawing an analogy with the scaremongers that decried the pocket calculator, prophesying it would degrade mathematical understanding. But, according to Stuart Russell, it has reduced our ability to do mental arithmetic. You might say that doesn’t matter, but we have to accept that there are trade-offs about which we need to think very carefully indeed. Convenience isn’t everything. Not knowing how to think could prove very inconvenient indeed.

The more time doctors spend staring at digital medical records, the less time they spend actually looking at their patients. The more time parents spend on parenting apps, the less they engage with their kids. The more time we spend with people that we’re predicted and programmed to like, the less we connect with people who are different from ourselves. And the less compassion we need, the less compassion we have. What all of these technologies attempt to do is to force-fit a standardized model of a predictable reality onto a world that is infinitely surprising. What gets left out? Whatever can’t be measured: style, taste, discernment, intuition, thinking, love.

AI will require that we all make choices: about the capacity and skills we want to hone and those we will lose. What will yours be?

If You Want to Go Deeper

Read Stuart Russell’s eminently readable book HUMAN COMPATIBLE on the dangers and promises of artificial intelligence.

For the Love of It

I’m always puzzled that extremely tragic films, books, plays can nevertheless leave us feeling uplifted. Aristotle explained it as catharsis, but I’m starting to wonder if there is something else going on in dystopic drama. I’ve seen 3 deeply depressing films lately: No Other Choice, The Sound of Falling and Sirat. I put them in a category that also includes Of Gods and Men, King Lear and Waiting for Godot. But Godot and Lear have jokes (of a kind) and Gods and Men has one of the most astonishing music sequences in all cinema. No Other Choice, The Sound of Falling and Sirat have neither. So why am I glad to have seen both? Perhaps for no other reason that they show a willingness to face into reality without flinching. These days, honesty can be a form of courage.