If you like First Dates, you'll love This is Dating

26 February 2022

9:00 AM

26 February 2022

9:00 AM

This is Dating; DeepMind

Apple Podcasts, Spotify and various platforms

An Artificially Intelligent Guide to Love

BBC Radio 4

The tagline of This is Dating, a new podcast from across the pond, is ‘Come for the cringe, stay for the connection.’ This sums up the listening experience pretty well. If the prospect of eavesdropping on a series of strangers’ first dates sends a shiver down your spine (some of us have endured enough disastrous dates of our own), give it ten minutes and cupid’s arrow should slowly begin to sink in.

The concept is similar to that of First Dates, the reality TV show in which lonely hearts pair up for dinner and judgment while a sexy French maitre d’ looks on, pitying the lack of social skills on display before him. The difference is that This is Datingdoesn’t play for laughs. It is unusually sympathetic towards the singletons who agree to have their Zoom dates broadcast down the airwaves. The team of women behind the podcast, one of whom is a ‘behavioural scientist’, even sit in and place prompts and topics of discussion in the Zoom chat box so the conversation never runs dry. It’s all very 2020s.

Virginia and Nick don’t hit it off straightaway. ‘What is the first thing you notice about a person when you meet them?’ she asks. ‘Their smile,’ he replies. ‘Me too. And their hair.’ ‘I don’t have a lot of that, so…’ They are looking at each other down the screen. Nick, sipping on his Grand Marnier, says he isn’t into bondage. Virginia describes a visit to a sex dungeon. How they get to a second date is anyone’s guess.

In another episode, a woman describes the time she agreed to go for dinner with a stranger only to find herself taking him to a drive-through and paying for the hundred tacos he insisted on ordering because he had ‘forgotten’ his wallet.

It could get tedious, but the presenters’ commentary and psychoanalysis are layered over each date, interrupting the proceedings at just the right intervals. The production is truly pristine. I may just have fallen under its spell.

Given some of the stories, one has to wonder: would we be better off dating machines? Listening to An Artificially Intelligent Guide to Love, a Radio 4 drama written using an algorithm, I’m not so sure. The play is promoted as ‘experimental’, and clearly aspires to an Eimear McBride kind of stream of consciousness, but the result is horribly monotonous. ‘You’re amazing. You’re so stunning. You’re beautiful. You’re so amazing,’ it goes. ‘The algorithm is glitching on compliments,’ a human collaborator on the play explains.

The algorithm, played by Fiona Shaw, is supposed to produce a guide to love. Its poetry isn’t bad (is that Sappho lingering in the lines of a lover being observed and inflamed by another?), but storytelling is certainly not its forte. The human uses the machine to narrate episodes from her life. We hear of her dates with various women; of her parents’ marriage via a bizarre episode involving a dead cockchafer; of a man who drops such unlikely lines as, ‘I want to get better at marketing.’ Even with Shaw in the leading role an hour of this was too much.

The play nevertheless raises some interesting questions. What is the difference between going on a date with a human and having them fill in the gaps of conversation as amenably as possible, and hearing an algorithm do the same? Is one more creative than the other? Has the algorithm the advantage in its ability to surprise?

AI is certainly becoming more sophisticated. Mathematician and all-round excellent communicator Hannah Fry is back with a second series of DeepMind, a splendid podcast that explores the development of artificial intelligence and robots and their application to daily life. In a recent episode, ‘Let’s Get Physical’, she walked us through a lab in London’s King’s Cross, where a series of booths with privacy screens summoned the image of a hospital. In one lay a disturbing range of motionless mechanical arms.

Teaching a robotic hand to insert a USB stick into a computer, or to turn a lock, turns out to be much like teaching a dog its first tricks. Where a puppy receives a treat for learning to sit or fetch, the algorithm is allocated points for getting something right in what is known as ‘reinforcement learning’. A robot can apparently be taught to walk in just 24 hours. Impressive, though the more I heard, the more I came to appreciate the complexity of human emotion and the impossibility of replicating it so precisely in a lab. Pinning it down outside the virtual world is difficult enough.

Got something to add? Join the discussion and comment below.

You might disagree with half of it, but you’ll enjoy reading all of it. Try your first 10 weeks for just $10

Show comments