Embracing AI at School and at Work

While Google has always allowed us to find answers to pretty much anything, it was only ever a repository of human opinion, a place to find what we had previously said, written or filmed.

Generative AI is different. The internet has transformed itself from being a giant filing cabinet to a giant brain. We now all have a PhD in every subject.

This raises the question of what we do with education. With this generative AI, much of what we would have had to create ourselves as humans can now be done for us and, within the academic realm in particular, the results are often better than what most humans could produce.

History has shown us that technology always prevails and those who embrace it early and quickly usually usurp people and organisations who resist the tech tide.

Outsourcing Heavy Lifting

Chat GPT is, of course, front and centre. While many schools and universities have been trying to uncover or ban “ChatBot cheating”, one of the better approaches I’ve seen has been that taken at the University of Sydney.

First-year medical students taking the course “Contemporary Medical Challenges” have incorporated ChatGPT into the course.

Students were assigned a mission: to formulate a question on a modern medical challenge of their choosing, prompt ChatGPT to “pen” an essay on the topic, and meticulously review and edit the AI's output. They had to complete a minimum of four drafts and reviews, edit and re-prompt the AI, and then refine it into a submission-worthy final draft.

The main criteria for success was being able to manipulate the questions for ChatGPT to not only produce an optimal essay but to observe the process and the thinking they went through while editing the essay and how they re-prompted the AI to delve into the appropriate arenas of knowledge.

“We want to make sure the grads are not just getting ChatGPT to do their work, we want them to have discerning judgment, and a curiosity about the future,” course coordinator Martin Brown told the Sydney Morning Herald.

“You have to work with it. You can’t ban it – it would be crazy to.”

This is truly an enlightened approach.

It’s clear that there are different types of knowledge. We have basic memorisation, reproducing information, and collating information – something which educational institutions have traded in for centuries. But when AIs like ChatGPT can do that for anyone, for free, it’s time to revise education.

It might seem like a crazy thing to say, but the reason we don’t judge students on their ability to lift heavy things is that machines were invented before the modern K-12 school system was. Even when we do physical education at senior school, it really becomes all about human energy systems and biomechanics.

While it’s crucial we have this discussion on what education looks like now, it isn’t just a quandary we face for future generations. It will have a massive economic impact in Australia. Education is one of our biggest exports – even in a post-pandemic environment. If organisations collectively decide AI prompt engineering is more valuable than an undergraduate degree, we could see a music industry-style decline in a vital market for Australia’s economy.

Age Gates and AI

While we’ve had calculators and spell-checking computation for a very long time, no one would argue that those who know how to add, spell and write have a distinct economic advantage in life. There is a reason we only introduce calculators and computers into education once kids know how to read and write – we need to be able to judge the output.

We need to know what good looks like, even when a tool can transform what would usually be good into something great. We'll need to be very cautious when introducing AI in the early years of education and very deliberate about including it at senior and tertiary levels.

AI won't change the need for deep domain knowledge; in fact, it may exacerbate it. Those who know more will be able to extract more from the generative AI systems at our disposal. They'll know what to ask it, how to get better revisions and, most importantly, discern if what it has generated is acceptable. In some ways, AI will turn much of what we do into curating and conducting possibilities. And just as it does in the art world, this requires judgment and even taste.

AI in the Office

I’m old enough to remember when Internet Explorer arrived on the Microsoft Office suite, and how many companies banned the use of the internet in the office. Of course, it now seems laughable. And you’d expect we’d have learned from this folly. Yet the list of firms banning ChatGPT isn’t just long, it’s also prestigious. It includes Apple, JPMorgan Chase, Deutsche Bank, Verizon, Northrop Grumman, Samsung, Amazon and Accenture, among others.

If you’re wondering why most of them have banned it, they have all cited various concerns around data usage and privacy. It seems ironic, given that individuals and corporations generally don't tend to care that much in every other digital realm. I have no doubt that this approach, if it isn’t already, will seem more ridiculous as time passes. I’m also sceptical that any employee of these firms wouldn't be using it on their phone to help with their work.

What is clear is that there is an inordinate number of use cases for generative AI in any office or corporate setting. Simple emails, creative briefs, HR documentation, legal work, software code, strategy templates, project management plans, you name it. If words or visuals are part of a corporate role, then not using generative AI is akin to racing a car on two legs. It isn’t just ignorant; it’s going to be at the cost of productivity and impact shareholders. A better approach is to set some data and usage boundaries to alleviate the corporate sensitivity concerns.