Through its website and daily newsletters, The Peak (a ZoomerMedia property) offers Canadians the news they need to understand business, tech, and other must-know stories.
In this dispatch, The Peak looks at the benefits and dangers of AI, from how Canadian towns are using the tech to their advantage to how AI-powered images and information are blurring the line between reality and illusion.
How Canadian Towns Are Using AI
AI could already – quite literally – be hitting the streets in your town.
Driving the news: Edmonton-based Runwithit Synthetics secured $3.5 million in venture funding from Raven Indigenous Capital Partners to grow its synthetic twin technology, which lets cities combine their data to simulate things like disaster responses, infrastructure investments, or the impacts of climate change.
Why it matters: Even as higher levels of government grapple with implementing AI, municipalities are getting more comfortable putting it to work on day-to-day tasks.
- Ottawa is using AI to predict homelessness, following a similar effort in London, Ontario. Data like age, gender, previous shelter usage, weather, and economic indicators are used to predict how often someone will use a shelter over six months.
- Saskatoon, Saskatchewan, predicts when buses need maintenance. The city didn’t have a service disruption due to a bus shortage for the year following AI adoption across the fleet in 2022.
- Leduc, Alberta, scans organic waste bins for contaminants, like plastic. First adopted last year, the town extended the project to the end of 2024.
- Cityrover’s machine vision detects potholes and other road maintenance needs, and has been used in cities including Winnipeg and Ontario’s Richmond Hill, Windsor, and Markham to bring down the cost and time for repairs.
Big picture: Given the questions around AI’s reliability, towns have been going slow, testing it on a case-by-case basis. But Edmonton created an AI framework so all of its data is organized and ready for AI opportunities that arise, from big urban planning projects to directing the daily work of building inspectors.
Yes, but: Cities have a lot of data on residents, and high-impact AI is largely unregulated until Bill C-27 is passed. For things like homelessness, that means potentially running into AI’s well-documented bias issues. But on a smaller scale, residents might not be thrilled with a robot rooting through their trash. — Josh Kolm
AI Might Mess With Our Concept of Reality
Google’s new AI photo tools are scary good. Pixel 9 phones will let users make all kinds of AI-powered alterations to images, from inserting themselves to adding or removing things with just a text prompt. Doctoring photos has existed about as long as photography, but it no longer requires technical know-how to pull off.
It will shift the burden of proof for truth. At The Verge, Sarah Jeong writes that a big problem with this is not just the stream of altered images we’ll be peppered with — it’s that people are going to assume pictures are fake from the get-go. It’s never a bad idea to be wary of what you see online, but it’s up to skeptics to find evidence to back up claims that something’s fake. Soon, people may have to be the ones to prove their pictures — from a dinged car after a fender-bender to a politician doing something they shouldn’t — are legit.
And it’s not just photos. A trailer for Francis Ford Coppola’s Megalopolis quoted negative reviews for the filmmaker’s classics to stick it to the haters, but the trailer was pulled after it was discovered the quotes were made up. There’s no proof that the quotes came from ChatGPT, but that quickly became the conclusion people jumped to. — JK
Why Nvidia Invested in a Canadian Deepfake App
Is an AI app that makes you look pregnant while shaking hands with Donald Trump the future of entertainment?
What happened: Wombo, a Toronto-based generative AI startup, secured a US$9 million dollar funding round. Led by Round13 Digital Asset Fund, the round also had participation from chip giant Nvidia and AI infrastructure startup CoreWeave.
Catch-up: Wombo doesn’t use the word deepfake, but the Wombo Meme app fits the definition: AI-generated images of celebrities and public figures. Right now it is focused on the U.S. election, promoting images of a cleavage-showing Kamala Harris or a muscular, shirtless Trump sitting on Elon Musk’s lap.
- Wombo got its start with a lip sync app that could turn still images — of yourself or others — into videos singing along to a song.
- The company’s other app, Wombo Dream, is a more straightforward AI image generator, creating images in different art styles with a text prompt.
Why it matters: Deepfakes and disinformation are a big concern in AI. But as OpenAI and Google put guardrails around generating images of real people or deceptive election content, Wombo proudly touts its ability to make election memes from fake images.
What they’re saying: Ben-Zion Benkhin, Wombo’s CEO, tells Peak Tech that the content on Wombo Meme are clearly jokes and users are tied to certain templates — by leaning into the silliness, Wombo maintains no one could mistake its content for a real image.
- We got Wombo Dream to generate images like Justin Trudeau taking a bribe from Drake and Donald Trump pointing a gun in the distance, but as Benkhin says, they do look obviously fake.
- Should pending laws limit the kind of content it can generate, Benkhin says the company would gladly stop and move on to something else.
Zoom out: In the short-term, Wombo has been using AI to chase viral moments, seemingly feeding into “AI slop” that blurs the line between spam and content. But long-term, Benkhin claims Wombo’s popularity is a proof point for where entertainment is heading — fully customized and personalized entertainment people create themselves. — JK
Visit The Peak’s website and sign up for the free daily email on Canadian business, tech, and finance that you’ll actually enjoy reading.
RELATED:
The Artificial Intelligence Boom: Understanding the Hype (and Concern) Over ChatGPT
