Giving homework to children may soon be a thing of the past, but at what cost?

ChatGPT, the cutting-edge AI language model developed by Open AI, is revolutionizing the way we interact with technology. I did not write this sentence – Chat GTP did, as I asked it to start a column on the subject for me.

I would never write such an opening line because a) it is dull, and b) it assumes you know what any of these terms mean. Doubtless, though, the revolution is upon us and the kids know all about it already and some are using it to write essays.

Let me explain. You simply sign up to an account for free, give the software a task, eg: “Write a biography of Harold Wilson”, and it will generate “human-like text” in response. It’s a sophisticated chatbot co-founded by Elon Musk. Within seconds it can write emails, answer questions, write essays and er… poetry. Too good to be true? Well, yes, at the moment. But Google is very worried about this development, as well as it might be, as ChatGPT not only searches the internet, but also rates the results for you and comes up with a direct answer.

On a more mundane level, teachers are worried that pupils are using it to do their homework. The smarter ones take a GPT essay and alter it a little so it’s not pure plagiarism. Others don’t bother and teachers are being given work that they may not always realize is not that of the student. The New York City Board of Education banned it from schools last year. A professor in the Netherlands referred to what it produces as “a bull—–er on steroids”.

Now British schools are facing the same dilemma. It may seem very “20th century” to even call this cheating, but as long as we expect kids to regurgitate knowledge in their own words, it is.

Talking to friends whose children use ChatGPT, it’s clear that it works much better for science, technology and maths than it does for creative subjects because it lacks nuance, critical thinking and any conception of ethics. Its database is limited, ending in 2021, and it just gets facts wrong. But, undoubtedly, that will change. So will homework become redundant?

Actually, I wouldn’t mind if it did. Primary school kids are now getting homework at six. (In Finland, children haven’t even started school at this age.) The pandemic showed us that most learning can be done in a couple of hours a day, as most of school is about socialization and crowd control.

Children need to learn how to study independently, of course, but they are already using a lot of technology to do that. Still, not enough information is given to pupils about exactly what that technology does and what can be trusted. We need more than one source. Something off Wikipedia, YouTube or a quick Google is not enough. Some cross-checking is always necessary. Fake news has thrived because of a combination of digital media spreading untruths and a passive acceptance of “alternative facts”.

AI muddies the water here just as it does with visuals. All sorts of fantastic visual images can now be generated online by AI apps like Midjourney or Deviant Art but who owns these images? A group of visual artists are now suing such companies for copyright infringement. This is a minefield.

If AI can make art and poetry and write articles, will jobs be lost? The news website Buzzfeed, for instance, is using ChatGPT for “content creation” while cutting its workforce by 12 percent.

There is no point being a complete Luddite about this. There may be many creative possibilities here. But let’s be clear: there is an ever-widening gulf between those who will not use a self-service till or a cash machine and those who will happily embrace an untrustworthy chatbot to write “journalism”.

It will be students who push us to think about this. If there is one skill that an expensive degree equips a student with, it is surely the ability to research, write and structure an essay. If that is done by a machine what do we lose? What is the difference between ‘human-like’ writing and human writing? Can integrity ever be programmed algorithmically?

Not quite enough. If poetry is the pinnacle of writing – which it surely is – ChatGPT has a mountain to climb. I asked him to write a poem about me and it was robotically flattering, I guess. “She’s a voice for the underdogs/and always stands up for the rights of the lost and lost” went one line. Logs??

I don’t think that’s going to win any prizes, although, of course, everything lovely he said about me was the absolute truth.

The odd reality is that ChatGPT hasn’t done enough homework yet. When it does, the distinction between human and ‘human-like’ will be harder to maintain, but that is the future.

And it’s closer than you think.


Leave a Reply

Your email address will not be published. Required fields are marked *