Pop culture tells us that the birth of artificial intelligence will give rise to killer robots, and the only way to defeat them is through a hydraulic press, pits of molten metal, blackening the sky or nuclear charges.
Subscribe now for unlimited access.
or signup to continue reading
How wrong they were. Turns out, all you need to kill AI is privacy advocates, the owners of intellectual property, luddites preaching "the end of jobs" and lazy university lecturers.
I am, of course, talking about large language models like ChatGPT. It's the AI chatbot which can do everything from summarising reports and recommending restaurants to writing job applications, code and poems.
Pitted against real doctors, one study found that ChatGPT was not only more accurate in answering patient's questions, it was more empathetic, too.
In another study, an AI system was able to identify somebody's gender by scanning their retinas, something humans didn't even know was possible.
Two papers have found that it has increased the productivity of programmers by 50 per cent.
ChatGPT can give a recipe based on a photo of what's in your fridge, it can explain images to people who are visually impaired, it can create a website based on a hand-drawn sketch, code an entire computer game in 20 minutes and even help with match-making on dating apps.
Thank goodness it's here.
Australia's productivity growth in the last decade was the lowest it's been in half a century. If you want your kids to have a better standard of living than you've had, you should be very worried by that statistic.
Economists have been pleading with governments for decades to undertake reforms to lift productivity growth.
Alas, the reforms undertaken have been piecemeal, small by historical standards and nowhere near what is required to get out of our productivity slump.
Our only hope, it seems, is that technology saves the day. Right on cue, here it is.
ChatGPT is far from perfect. But it is improving very quickly. It has already showed its ability to undertake countless tasks, freeing up people's time so they can focus on higher-value activities. It is the definition of a productivity-enhancing technology that will boost wages.
And how have we responded to this opportunity?
Privacy advocates are trying to suffocate it in red tape. The owners of intellectual property want to tax it to death. The end-of-work luddites are trying to shut it down. Lecturers are trying to ban it.
These are different groups of people who are falling for the same fallacies.
The first is the free-lunch fallacy.
Economics is about trade-offs. Nothing comes for free. And the same is true for privacy and the protection of intellectual property rights.
Every group spoken to as part of the independent review into Australia's COVID response said that the lack of data made our programs less effective, made it harder to target services and cost lives.
READ MORE:
Privacy advocates extol the benefits of having more privacy, but ignore these lost lives.
Owners of intellectual property extol the benefits of patents and copyrights, but ignore the costs like reduced access to life-saving medicines.
We need to weigh-up costs and benefits, and not pretend that we can have a free lunch. The same is true for technologies like ChatGPT.
ChatGPT definitely raises issues for privacy and the infringement of intellectual property. Try asking it for your biography, phone number, email address and street address. You might be alarmed what's out there.
But whatever the costs, these costs need to be weighed up against the benefits of these new technologies, which brings us to the next fallacy.
The second fallacy is the lump of labour fallacy.
In the 1700s, the so-called Luddites tried to ban mechanised looms and knitting frames because they would put textile workers out of work.
We've seen similar backlashes against everything from cars and computers to robots and rail lines. But these fears have always been wrong.
The lump-of-labour fallacy is the false belief that there are X jobs available in the economy and if the number of people exceeds X then the result is unemployment.
This is not how the economy works. History shows that new technologies make labour more productive and create more jobs than they destroy.
Those fearing that ChatGPT will destroy more jobs than it creates or improves have not convincingly explained why this time will be different to all the other times they've made the same claim, and have been wrong.
After all, jobs rarely involve a single task. The replacing of some tasks by ChatGPT will free up people's time to focus on their other tasks. It's productivity improving.
The third fallacy is the appeal to tradition. University lecturers are trying to stop students from using ChatGPT to write essays and complete their assignments. Again, this is the wrong approach.
Lecturers should be encouraging students to embrace these new technologies and use examinations to test their ability to use them, to ask the right questions and to critically evaluate their outputs.
Lecturers and teachers need to adjust and stop appealing to tradition.
ChatGPT will raise a bunch of issues which we need to manage. But we need to be honest about the costs of regulation, approach it through a cost-benefit framework, sensibly manage risks and embrace opportunities so our kids can have a better standard of living than us.
Let's not kill the golden goose before its even laid its first egg.
- Adam Triggs is a visiting fellow at the ANU Crawford School and a non-resident fellow at the Brookings Institution.
