I don't care about robots ruling the world, I just want my house to clean itself!
Yes, there are apocalyptic warnings of a future war between humans and robots; but can AI get the dust from under my couch first?
January 20, 2025
Here’s a challenge: stop the nearest person on the street and ask them what AI stands for. Now keep doing that until you find someone who doesn’t know. Tired yet? That’s because you’ve been asking for two days and your family misses you.
We can’t avoid artificial intelligence chatter anywhere; television, work (especially work), dinner discussions with friends. What can and can’t it do? Was Will Smith’s I, Robot a thriller or a documentary? When will it inevitably rise up and lead us? For all of the comments about its “chance above zero to kill us all” and the fears of Canadian AI legend Geoffrey Hinton about “the end of people,” there are the Marc Andreesens (famed software engineer and venture capitalist) telling us “Why AI Will Save the World” (it’s literally the title of his piece).
But let’s leave the moguls and innovators to debate the fate of humanity for a moment, and ask the question that really matters:
When can I outsource my vacuuming?
Are we meant to believe that artificial intelligence is so special that it may literally be the most profound change any of us ever live through, but currently it would still be confused by that stain on the counter? Well, yes and no.
How many robots does it take to change a lightbulb (or put away a can of soup)?
Researchers from Carnegie Mellon University have actually determined a new way to train robots to help with chores, like opening a drawer, oven door and lid; taking a pot off the stove; and picking up a telephone, vegetable or can of soup. Where the true challenge comes in is teaching robots "context" (i.e. identifying colours or locating where certain objects specifically go). Yet with AI’s rapid improvement, is this just something that will be solved with time? Or could it remain forever beyond artificial capabilities?
Here in Canada, Sanctuary AI has developed a robot that has been packing clothes into plastic bags in the backroom of a Canadian shop. Yet it still struggles with context (for example, having to open the bag wide enough with one hand and place clothes inside with the other).
“I’m not telling it those are the wrong berries, you tell it!”
Even robots have to watch training videos at work
The solution, according to many working to solve these issues, is proper video training, where the robot can watch a task (or thousands, or millions) being performed, to learn. Still, there are so many edge cases that may have been missed in training that a 100% success rate still feels a long ways away. With that said, the University of Oxford believes that by 2033, the average amount of time we spend on “a domestic task” will be reduced by 39%; and as crazy as it seems, there’s a belief that something that still struggles to properly open a bag can safely identify, collect and dispose of nuclear waste in our lifetime.
The future is unknown – but it will definitely be interesting
No matter how often we hear of AI’s exponential growth, it’s still almost impossible to wrap our minds around the reality of it. Beyond its ability to even work correctly, concerns abound about shortcuts that could lead to longer-term risks. Blair Radbourne, who leads the Enterprise Technology & Cybersecurity team within the broader Data & Technology department at OMERS, has some thoughts. As an executive with a proven track record in the Canadian financial services industry, Blair possesses extensive expertise in Cybersecurity, Infrastructure Architecture, Budget and Resource Planning, Stakeholder Relationships, and Technology Governance & Compliance.
“What happens if people's tendency to weigh purchases through a quality versus price lens leads manufacturers to build the cheapest AI/robotic cleaners by violating Isaac Asimov's "Three Laws of Robotics," and these devices will obey orders or protect their own existence as their highest priorities?” he says. “We’ve seen this movie so many times in the Internet of Things space, where basic security is left to the side so that the cheapest minimum viable product can hit Amazon the fastest.”
There’s a lot to take in here, and many angles to see it from. Yet one thing seems clear; no one is truly sure how the AI revolution will play out. If AI doesn’t advance as quickly as many believe it will, the detractors may be able to relax a little and back off its sci-fi implications. And if it does continue to progress rapidly, you can tell your great grandchildren about an old machine that you used to have to push around the house to pick up dirt, and days when you had to personally make your food before you ate it.
Sometimes, you used to feel downright robotic.
The Relatable Economist is an ongoing written series focused on how the economy, geopolitics, markets and more are impacting our day-to-day lives, discussing topics that matter to you, even if just to share with your friends at your next get-together or in the stands at your child’s or grandchild’s soccer game. Have a topic you want to learn more about? Write to us at therelatableeconomist@omers.com