Pages

Sunday, October 8, 2023

Michael Johnston: I'm the operator with my large language model


In 1981, the German electronic band Kraftwerk released a song called Pocket Calculator.

At the time, affordable calculators were a recent phenomenon. Kraftwerk, as dyed-in-the-wool technophiles, were clearly pleased.

Many educators were pleased too. Young children often don’t enjoy learning arithmetic. Those educators thought that calculators could free them from having to learn it. Instead, they could learn to apply mathematics to real world problems. This, they thought, would be more motivating.

Other educators were less pleased. They were concerned that, if children came to rely on calculators, they would never learn basic numeracy. Some were also worried that students might use them to cheat in exams.

More than forty years later, educators are having what is essentially the same argument all over again. Except, this time, it’s about writing. Chat GPT and its large-language compatriots can write flawless, if often pedestrian, prose.

Like pocket calculators before them, AI has mesmerised fans of technology and stimulated an important debate in educational circles. Excitable futurists imagine freeing students from the drudgery of learning to write grammatically correct sentences. Conservatives worry about falling educational standards and cheating.

So, which side has it right?

In the four decades between the first mass produced calculators and the advent of generative AI, the science of learning has advanced greatly. It can now help us to resolve these debates.

When we first encounter new knowledge or start to learn a new skill, we hold it in a short-term memory system called working memory while we consciously manipulate it. Working memory has a very small capacity. When it gets overloaded, we feel confused.

To manage this problem of cognitive overload, we must learn in stages. Knowledge must be committed to long-term memory, which has a virtually infinite capacity. That frees up working memory to take in new learning, building on what has been stored in long-term memory.

Mathematics and writing are both cases in point. To learn algebra, the rules of arithmetic must be committed to long term memory. To write thoughtfully or creatively, basic writing skills must be similarly mastered.

Just because technology can do something as well as, or better than, human beings, does not necessarily mean that children no longer benefit from learning it. Skipping the basics can close the door to rich and interesting learning.

Sure, Kraftwerk used computers to help them make music. But they learned to play well first.

Dr Michael Johnston has held academic positions at Victoria University of Wellington for the past ten years. He holds a PhD in Cognitive Psychology from the University of Melbourne. This article was published HERE

2 comments:

Gaynor said...

I have a passion for teaching the times tables by rote and I believe any arguement against this is nonsense. This staunch belief applies to arithmetic methods, as well.

My initial practical work in learning this almost lost art of times tables was in teaching my own son who was bright but had a memory like a sieve. It was totally tedious. I wanted to give up but remembered back in standard 6, everyone knew their tables perfectly.

In teaching my son I employed chanting with or without repetitious writing out, writing on his hand some facts to repeat all day, flash cards, charts, games, mnemonics, work sheets, tricks and other techniques along with motivational rewards.

I heard of an old time NZ school teacher who threatened his students with the strap for failing to know their tables within a reasonable time frame. Apparently this was very successful. Another school tactic was the use of the motto, 'No calculator until tables known.' A certificate of achievement was presented to the kid as well.

It is pleasing cognitive science has come along to rescue, previously condemned rote learning of some things particularly the basics and told us there is an actual section in the brain which stores facts and methods permanently. Before calculators, times table charts were used but common sense alone dictated , as for calculator s now that these destroyed automaticity in calculations.

Anonymous said...

Yes, Gaynor, you are absolutely on the money! Despite the current narrative (which is so obviously failing our young), the old tried and proven methods work.

When are parents going to wake up, take responsibility, and demand more from our Education Ministry that so clearly has lost its way?

Meantime, let's give, Rose Hipkins another gong. Yeah right!