L-R: Mairin Loewen (Associate Program Director, Urban Climate Leadership), Laura Ross (CUPE legal representative), Mark Hancock (CUPE National President), moderator Diane Therrien (CUPE research), Marie-Christine Therrien (Director, Cité-ID LivingLab).
Municipalities must think through what the adoption of artificial intelligence (AI) means for workers and the community, according to speakers on an expert panel presented by CUPE at the Federation of Canadian Municipalities conference this past weekend. Panelists discussed risks related to equity, data governance, job loss and climate change.

Marie-Christine Therrien, Director for the Cité-ID LivingLab, said municipal leaders should ask a few fundamental questions before adopting AI tools: What is the quality of and bias in your municipal data? Are staff trained on privacy and data governance? What is the back-up if the AI system fails?

CUPE legal representative Laura Ross highlighted that municipalities need to build trust with technology adoption. “If AI-based services aren’t working properly,” she said, “you risk degrading public services and the public’s trust in those services.”

Mairin Loewen, Associate Program Director of the Urban Climate Leadership project, stressed that large language model AI systems are extremely energy intensive. She recommended ensuring internal policies and strategy on AI reflect a climate lens.

Loewen urged municipal leaders to be informed and prepared for data centres that are being proposed in communities across the country to power AI’s growth. These data centres consume massive amounts of energy and water.

Ross raised concerns about AI algorithmic management systems which can fall prey to discriminatory bias, and encouraged municipal leaders to consult and negotiate with unions on transparency and accountability of AI systems, as well as on surveillance and monitoring, bias and discrimination, and data management.

Download CUPE’s resource for municipal leaders: Digital technology and AI in municipalities: Important questions.