Friday 24 February 2017 7:30 am

My best friend the algorithm is a better human than you are

Thanks to my friends, I discover some outstanding TV drama from time to time. Like Lost, for example, or the first three seasons of Homeland.

Two years ago, someone suggested that I watch a show. To be precise, they did not suggest, they insisted – indeed, demanded – sending me message after message which said: this TV drama is for you, watch it, you won’t regret it.

The drama was The Bridge, and it was indeed for me: I have since watched The Bridge three times, and I will most probably do it again. And the “friend” that suggested it was Netflix, whose algorithm seemed to know me better than any of my human friends.

We have all heard of automation, and the risk that it poses to jobs. The common understanding is that the jobs that are under threat are low-skilled and mechanical, whereas those jobs that require human intelligence are safe.

In many cases, the opposite is true. Last year, my local Tesco installed automatic check-out scanners, and at the time all the cashiers were worried about losing their jobs. A year later, they are still there – only now their main job seems to be trying to trick the scanning machines into recognising the bar code. The only person likely to lose his job in this situation is someone at the Tesco headquarters.

On the other hand, I can name a few activities traditionally associated with the unique capabilities of the human brain which could be easily replaced by automation.

President Trump’s morning tweets, for example. Their algorithm is extremely simple. At 7am, search the internet to check who has insulted you most in the past eight hours. Use the following search criteria: “charlatan”, “bully”, “pathological liar”, “tangerine”. Decide on your retaliation target. Start drafting, using the following composition methods: block caps, exclamation marks, primitive syntax. Lie blatantly about something. Finish with a catchphrase. Post.

The other day, my editor took a bit long to publish my column. Using the Trump Twitter algorithm, I could programme the machine to tweet the following: “VERY, VERY badly treated by the so-called features editor of @CityAM. Millions of people read my columns. Enemy of the people! #ColumnistIsForLifeNotJustForChristmas”.

Another product associated with unique human ability – but which could be easily automated – are interim results press releases. Here, the algorithm is also simple. At the beginning, include a few numbers. If numbers are bad, blame on market conditions or currency fluctuations. Mention track record. Talk about cost control, and how it has improved. In the chief executive quote, say she is delighted and looking forward to something – does not matter what. Send to everyone in your media database, even those who write about food or horse racing.

And finally, to preserve the sanity of the populace, most customer services jobs need to be automated as soon as possible. I would rather press buttons on the phone than spend 30 minutes with someone who is trained to only read from a script. Given the appalling standard of most call centres, the machine could even return a better result – and it would certainly spare a shouting match and a need to watch something silly on Netflix afterwards to calm down.

In the meantime, I eagerly await the next recommendation from Netflix. In fact, I actively co-operate. When I am not around, I leave Netflix to play films that I had seen in the past and liked. That way, the algorithm should become even more precise and attuned to my tastes. So far Netflix has not got in touch. Which just shows its genius: it knows that nothing can match The Bridge.