The most immediately interesting to me about the fascinating talk on future trends at Microsoft’s Future Decoded conference in London was not the future at all, but the present.
Apparently, the text recognition capabilities of artificial intelligence (AI) is now equivalent to that of the average person. Ditto image recognition and, appealingly, text comprehension.
Now, I’ve always been interested in the use of technology to alleviate the workload associated with marking, but have never been fully satisfied with the (affordable) solutions on offer when it comes to marking short answers or essays. For example, with one product, you had to enter the correct short answer, which might be:
“This is an example of an If-Then-Else statement.”
If a pupil entered the answer:
“This is an example of If-Then-Else”
the answer would be deemed to be incorrect, which is clearly not the case. If the computer program was able to understand what the pupil had entered, as opposed to simply matching it against what the teacher had entered as being the correct answer, not only would the marking be more accurate but the teacher could be more ambitious in terms of the questions she sets for the students.
The only issue is that, given how the AI learns what the text means (by being fed lots of examples of whatever it is that it is being taught), any student who comes up with a startling new theory is bound to marked down by an algorithm which doesn’t recognise it as such. I have not seen this discussed anywhere except by myself, and having a discussion with oneself is never a completely satisfying experience.
This is bad news for any teacher wishing to delegate all of her marking to AI, but perhaps a source of some solace for anyone worried about the robots taking over.
I hope to write more about the conference in my newsletter, Digital Education. To subscribe to it, please go to this page: Digital Education.