There’s been trouble at'mill, recently.
For those of you who live outside these shores, our young bright things have in recent weeks been receiving academic qualifications that, in any other year, would reflect exams they would have sat in the early summer. Covid-19, of course, carried off our exam season alongside our elderly, but the whippersnappers still deserve an envelope of letters.
So what those who know best did was put teachers' assessments through an algorithm (a word many, many more Britons now know than they did in July) that took into account factors such as teachers maybe wanting their charges to do their best and being a bit too glass half full.
Quite why baffles me, but those powers decided to then publish both the teachers' assessments and the revised (or, if you read the press, 'downgraded') markings. Cue much wailing and gnashing of teeth, followed by embarrassed back-tracking so that teachers' grades prevailed. (It's actually worse than that - kids get the higher of the processed grading or whatever Mister Chips thought of first). We’ve even had the brilliant life-imitates-art-imitates-life story of a girl who wrote a story about academic grades being decided by an algorithm being upgraded after being initially downgraded by the algorithm.
This has particularly piqued my interest, as it straddles both my professional life in human resources, where bringing consistency to performance assessment within one company, let alone one nation, is an unwinnable war, and science fiction. I was even briefly a ministerial speechwriter during the Blair years focused on performance management in the teaching profession.
Teachers are brilliant at teaching. If I say anything different, I'm liable to be chased by a mob and strung up from a lamppost ("The NHS is crap! How could I possibly be making it worse?!"). But what teachers are not in a good position to judge is how their assessments compare nationally, let alone to the school down the road. There needs to be some calibration, some standard setting, some way of adjusting for the schools who consistently over- or under-estimate.
Of course, the powers that be got the algorithm wrong. I'm not about to defend how they did it, but it needed to be done. Relying on teachers' assessments alone has had the effect of devaluing the nation's academic qualifications (pass rates at 16 have jumped from 70% to 79%; A-grade equivalents by a quarter), and completely shafting the higher education sector, who are obliged to take in everyone who successfully achieved the bar set for them, a bar set when pass rates were expected to be 'normal'.
I also sympathise with teachers. If this happens again, would I want to know that little Johnny Flick-knife's grades are solely down to me rather than me being one of several arbiters in this? Probably not. We separate teachers and examiners for the same reason we separate out judges and juries.
But what fascinates me is how the country has reacted to the idea of an algorithm. We may as well have called it Jimmy Saville. And that's where the sci-fi comes in. Because an algorithm is a sci-fi thing, a dark, malevolent, invisible force that can never mean good news. It's Orwellian, Big Brother-y. It's decisions being made about us, things being done to us, without us even being in the room. It not just us not having a say, it's us not even being aware that we're being talked about.
But we shouldn't be scared of algorithms (and, if you are, you're gonna simply shit yourself inside out when you meet the T-1000-meets-Alien-meets-Freddy Krueger bitch that is the heuristic). An algorithm is simply a set of rules. If then. That's all. Yes, it can be used for ill, but it's a tool, and like any tool, you can use it to bang a nail into a wall or a nun into a coma (not sure why I grabbed hold of that image, but anyway).
Heuristics save lives. Take the Goldman Index. It's an algorithm to decide on the small matter of whether you're dying. Or you could get a doctor's opinion. Put your money on the former if you want to maximise the chances of living. Why? Because doctors bring a whole lot of baggage that's inevitable with messy human thinking. It's this messiness that we mistake for sophistication, that makes us believe that 'everything has been taken into consideration' and that can only be a good thing. To quote Malcolm Gladwill's 'Blink', which is where I first read about it, “Extra information is more than useless; it’s harmful; it confuses the issues. What screws up doctors when they are trying to predict heart attacks is that they take too much information into account.”
You want another? Heuristics are better at making trades on the markets than humans. The linked article concludes, humans will be overseeing and validating what the machine is doing, but perhaps the lesson from the Goldman Index is that humans shouldn't have the right to veto if only the algorithm sees what truly matters and filters out the noise. But would the masters of the universe allow that? Aren't we humans made in the image of God? We create the machines, so how could they be more powerful than us? Go figure.
We need a bedding in period to accept the machines, even when screwed to the floor, even as equals - a deeply unsettling idea - which is what my Wall Street story 'The Thirteenth Floor' is about, published in Third Flatiron's 'Gotta Wear Eclipse Glasses'. Go and support a small press by buying it.
And, when you've read that, here's another two dozen.
No comments:
Post a Comment