Monday, 9 March 2026

There are no short-cuts in being human

This is an interesting thread (especially if you include the comments of others) on what happened in one lecturer's class when (she strongly suspects) her students began using AIs to summarise the study material that they were meant to read.

https://xcancel.com/Sally_Sharif1/status/2030403451663114603#m

There are no shortcuts from the back-and-forth of reading something with the desire to learn through interacting with it.

Unfortunately, in Western education for some time now, much of the process is often skewed by the idea that the important thing is to gain the qualification. The qualification was meant to measure whether the education had been gained; but as with other measures, is subject to Goodhart's Law, "When a measure becomes a target, it ceases to be a good measure" - https://en.wikipedia.org/wiki/Goodhart%27s_law . When people are focussing upon the piece of paper that says they're educated (or that the school they're running educates lots of people well, etc.), then actually becoming educated starts to slide, and slide, and slide. I wonder if it's even possible to hit a lower nadir than in the Covid years, when the UK government handed out grades and certificates which ultimately expressed "here's what we believe you could have got in an exam if we'd educated you".... and the merry system happily trundled on, undisturbed by the important realities that it was orginally set up to accomplish.

Current AI (based on Large Language Models), like many other things, is, as we experiment with it and discover its strengths and weaknesses, pointing us to the wonders of humanity, made in the image of God. Yes, perhaps you can get a machine (after much human study and ingenuity to make the machine) which can approximate and outwardly resemble the product of actual human thought. But that, in the things that matter, is generally unlikely to be of real use, and may cause real harm if I pretend otherwise. If an AI summarises a text for me, that could be useful if I wanted to check if I might have missed something from a mental summary that I had from having done the proper engagement and reflection through reading and thinking. But if I use it to replace the engagement, then I only get the mirage of learning, the self-delusion that I took part in learning..... and not the actual learning itself.

No doubt in some scenarios, a quick-and-cheerful summary of something straightforward has value, and the AI can do something useful there. (Note too, though, that such cases are likely to be the ones that humans were already doing well - you can probably easily Google for a human-generated summary, and the AI is likely good at the task because it ingested so many human summaries in its construction). But if it's being deployed to do something that is trying to imitate human thought, reflection, moral wisdom, creativity, and such things that are tied up with what it means to be a divine image bearer.... then it's probably being mis-deployed, and the results will be worse than if we had simply ignored it. Christian wisdom requires understanding the wider purpose and point of something within the Creator's overall plan and design.... not simply how to economise on the use of time to achieve higher efficiency, as modern Westerners understand efficiency. Modern Western culture is obsessed with superficial efficiency in order to increase economic output; the cost of this is paid elsewhere, and is significant.

No comments: