Toronto beach water quality predictions criticised as inaccurate

Released: June 2022 

An AI-powered predictive modelling system used by Toronto Public Health (TPH) to forecast water quality at two beaches came under fire for being inaccurate, unreliable, and opaque. 

Adopted by the Toronto authority early summer 2022, the AI predictive modelling (AIPM) system used a series of calculations based on historical data and metrics such as rainfall, temperature and wind direction. It also pulled real-time meteorological and hydrological data. 

However, it quickly came under pressure from water advocacy group Swim Drink Fish for repeatedly allowing beaches that tested high for E.coli using traditional means to remain open.

Per the Toronto Star, TPH responded by saying 'While AIPM is not expected to be 100 per cent accurate in assessing water quality, it presents a significant improvement over test results using the traditional means for assessing microbial water quality.'

Swim Drink Fish argued greater transparency about how the system worked was required from the Toronto Health Board if the model was to remain so that beachgoers could make educated decisions regarding their own health and safety. 

According to The Information, TPH officials 'did not respond to a question about whether officials ever overrode the model’s forecast. But data published by the city show that the posted swimming flags at the two beaches never differed from the model’s predictions.'

Operator: Toronto Public Health
Developer: Cann Forecast
Country: Canada
Sector: Govt - health
Purpose: Predict water quality
Technology: Prediction algorithm
Issue: Accuracy/reliability; Safety
Transparency: Governance; Black box

Research, advocacy 🧮

News, commentary, analysis 🗞️

Page info
Type: Incident
Published: December 2022
Last updated: May 2024