Making Sense.

We covered Vancouver's pushback on its planned AI data centres in previous issues, and it has now reached City Hall. The city is home to two of the three data centres Telus is building with the federal government and developer Westbank, as part of a cluster set to scale to more than 150 megawatts by 2032. When we last wrote about this, hundreds had marched on City Hall in late June. Now, organizers have a further rally planned for July 26.

To address this, Councillor Lucy Maloney has raised a motion, "Addressing the Risks of Digital Infrastructure," which went to the Standing Committee on City Finance and Services on July 15. It directs staff to report back on a public impact requirement covering water, emissions, noise, and heat, as well as a data-centre-specific business licence requiring annual disclosure of water and power use, and options for securing community benefit. The pause is a matter of sequencing, since the motion asks for the above work to happen before Council rezones 111 East 5th Avenue or permits 150 West Georgia.

Underneath the politics, BC also has a physical constraint. BC Hydro is running a competitive process that caps data centre power at 300 megawatts over two years, so projects here compete for electricity. In contrast, north of Edmonton, Meta broke ground last week on the country's largest data centre which will draw 1 gigawatt, which is ~60% the amount of power fuelling the entire city of Edmonton (and more than 6x what the entire Telus cluster is projected to need by 2032).

On July 9, DIGITAL, Canada's digital-technology innovation cluster, backed two mining projects worth $19.9 million, and one of them is led out of Vancouver. Koonkie Canada received $2.89 million for an AI system that tracks how well a mine site recovers once the work is done, by combining environmental DNA analysis, soil-health mapping, satellite sensing, and Indigenous ecological knowledge. Its partners expect it to shorten land-restoration timelines by five to ten years and cut costs by up to 40 percent. The second project, from Ontario's Novamera, uses subsurface imaging and AI-guided robotics to extract ore more precisely, cutting waste by as much as 90 percent.

This is AI adoption making heavy industry cleaner, cheaper, and less wasteful. Cutting up to a decade off a restoration timeline, up to 40 percent off costs, and as much as 90 percent less mining waste are not small efficiencies. They are the kind of gains that change what is economically and environmentally possible within the physical, resource-heavy work that much of this province runs on.

This month the Toronto-based Vector Institute released a free, open-source tool called UnBias-Plus, built to catch biased language in both ordinary writing and the data used to train AI models. It scans for bias around race, gender, age, and political framing, flags what it finds, explains why, and suggests more neutral wording. Since large language models learn from human-written text, they pick up human biases. When those quietly surface in things like hiring screens or clinical notes, they can cause real problems.

This is worth noting as AI starts making and shaping decisions about people. Bias in a model is not only an ethical issue, it is a legal and reputational one, and Canada's national AI strategy names it as a challenge. Now a free, open tool from a respected Canadian institute works to reduce that obstacle, meaning the ability to use AI responsibly is increasingly within reach of any organization.