Last updated March 26, 2020 at 8:28 am
The artificial intelligence future is inevitable, but what it looks like is up to human decisions – we need to make sure we get it right before we roll it out on a larger scale.
Why This Matters: It’s up to us to determine what the future of AI looks like.
From our phones to our toasters, the technology that surrounds us is increasingly turning ‘smart’ with artificial intelligence.
By 2023, just three years away, the global spending on artificial intelligence systems is estimated to more than double, hitting almost $98 billion. And with the revolution comes concern about just how much of a role it will play in our lives
According to Peter Eklund, a Professor in Artificial Intelligence and Machine Learning from the Deakin University’s School of Information Technology, we already rely on AI, like GPS and home devices through the Internet of Things.
“We already see a flavour of AI with Siri and Google Home and we’ll see more and more of these Internet of Things devices and we’ll need to be able to control and interact with them,” he says.
“Automatic Speech recognition in Google Home, Siri and Alexa was the frontier of AI just a short while ago, but now it’s commoditised and we take it for granted.”
The artificial intelligence future in reality
AI has the potential to automate routine and tedious tasks that would make any human shudder. Think data entry and compiling massive amounts of information into organised lists.
“We’ll begin to see automated algorithms in the workplace. Algorithms will start to determine the optimal sequence of tasks we perform, particularly in white-collar jobs. We’ve seen this already to a certain degree,” Eklund says.
“Imagine I’m a school principal and I want to get a complete list of all the students in the school sorted by their age but I only have individual spreadsheets as class-lists. You could imagine asking an AI to merge and sort these files, because that would be a really tedious task to do by hand, and if AI can provide those kinds of solutions then it’s a really big efficiency boost.”
A common argument against AI is that it will replace human jobs. However, Eklund says that won’t always be the case.
“AI will never renovate your bathroom, and I’m yet to be convinced it can entertain people, I seriously doubt AI will be able to drive the weekly bin collection trucks for instance,” he says.
Instead, he suggests that AI will contribute to low-skilled job industries in the way of safety cams, visual warnings and alerts via augmented reality.
“I don’t see, particularly in the physical workplace, workers being replaced by AI and robotics.”
Robodebt – when automation goes bad
Although AI sounds very nice and efficient, as with any new technology, there have been instances of it going badly wrong.
Cast your mind back to 2016 when the Centrelink Robodebt scandal made headlines nationally.
In that situation, the Federal Government-operated social services provider Centrelink used a computer program to gather data from other government agencies, like the Taxation Office, and then compared it with what people reported to Centrelink. The program was designed to quickly check whether there are discrepancies in what you reported to Centrelink and what an employer reported.
Those who were found to have a discrepancy were notified and instructed to check and explain the discrepancies.
If the discrepancy wasn’t explained, it was assumed to be an overpayment and a debt was issued.
The problem with the program was, more than two-thirds of the notices given needed to be re-assessed because they weren’t accurate and in many cases completely incorrect.
That led to people – even those in the process of contesting the debt – having to go onto a payment plan for a debt they didn’t owe, and caused serious distress.
“This is what happens when you get it wrong – the algorithm was wrong, and the scalability of the automation compounds the error to geometrically, way out of proportion to what could have ever happened if this had been processed manually,” Eklund says.
“The result is a legal class action. The government looks really mean. The department looks incompetent and the public loses faith in everyone.”
What stands in the way of AI?
Mistakes aside, AI continues to find its way into our daily lives.
This month alone, supermarket giant Woolworths announced a trial using AI to scan and automatically detect fruit and vegetable purchases.
In a fast-paced world, where efficiency and automation are all the rage, it’s easy to get swept away in the utopian wave of AI, but Eklund suggests that it’s important for us to temper our ambitions, especially before we roll them out on a larger scale.
“Somebody at Centrelink should have said ‘let’s slow down a little here. Maybe we’ll try this for 50,000 cases and see if we get it right’, then they should have rolled it out in a much larger way once it’s been fully tested.”
“There are so many lessons in shows like Black Mirror about a dystopian view of the AI-enabled near future. It sends a real warning to us about what we definitely don’t want from AI.”
According to Eklund, when it comes down to it, the only thing standing in the way of AI being successful is us.
“I think people stand in the way, because ultimately the algorithms are being written people.”
And despite the power of the technology, at the end of the day it’s people who will ultimately determine what the AI future looks like. Balancing those decisions will make sure we get it right.