Toronto seeks $35 million to make AI a permanent tool against congestion
The city is asking council for $35.4 million to expand AI at key signals, prioritizing suburban arterials and the Finch West LRT while keeping humans in the loop.
Copy link
By Torontoer Staff
Toronto’s transportation services is asking council for $35.4 million to permanently expand the use of artificial intelligence at traffic signals and intersections, with Mayor Olivia Chow’s budget endorsing the increase. The proposal targets major suburban corridors and includes a priority program to improve Finch West LRT operations.
Officials say AI can respond to changing conditions faster than manual control, freeing control centre staff to focus on constrained downtown corridors and other complex situations that still require human judgment.
What the funding would buy
Budget documents show the $35.4 million would expand AI-based signal control across Scarborough, North York and Etobicoke. The funds would pay for software, rooftop corridor cameras to give operators a broader view than single-intersection feeds, pilot implementations and contracts with technology vendors. The city says further funding requests will follow as the program scales.
How the AI would work on Toronto streets
City staff would set desired outcomes for specific intersections at particular times, such as reducing queue length or improving throughput, and the AI would make real-time adjustments. Those adjustments can include extending green phases, shortening red intervals or coordinating signals across a corridor to smooth flow when demand spikes.
We have anywhere from three to five people in the traffic control centre at any given time manually intervening in traffic issues, and Toronto’s got more than 2,500 traffic signals. You do the math.
Roger Browne, director of traffic management
Currently staff can monitor about 300 intersections on screens and often react manually. The proposed AI would respond instantly to incidents that push drivers onto alternate routes, reducing the time humans spend making incremental adjustments.
Where AI makes sense, and where it does not
City officials plan to prioritise arterials and corridors with longer link distances and fewer constraints. Officials say downtown streets, with short distances between signals, many turn prohibitions and one-lane streets, offer limited opportunity for autonomous adjustments without creating knock-on effects.
The downtown core is already so very heavily constrained. So the decisions it can make are very limited. It can’t really be creative.
Roger Browne
- Suburban corridors with multiple lanes and longer spacing between signals are priority targets.
- Downtown will remain under closer human oversight to avoid unintended chain reactions.
- Rooftop cameras will extend visibility beyond single intersections for better corridor-level control.
Finch West LRT is top priority
The city identifies the Finch West LRT as the top operational priority. AI would detect trains and ensure they clear intersections before allowing conflicting left turns, and provide data to re-time signals if the TTC can increase train speeds. Current transit signal priority only activates when vehicles are running late.
We can detect the train and make sure that the train clears the intersection before the left turns go.
Roger Browne
Report cards, accuracy and potential blind spots
Part of the plan is to generate regular 'report cards' that show signal performance across corridors multiple times a year, rather than the current annual reviews. Vendors that have worked with the city say those reports can highlight recurring green-time shortages and coordination failures.
The report card fundamentally looks at what’s happening at a given intersection and then in the broader network. Maybe in the afternoon, everything’s fine but in the mornings you have a whole bunch of failures.
Kurtis McBride, CEO of Miovision
Experts caution that AI must be fed the right inputs and rules. If software does not account for weather, for example, snow conditions could change priorities for pedestrians and vehicles. There are also questions about how to measure fairness across users at an intersection.
How reliable is it that the AI is actually aware of who is waiting and how many people are waiting? AI simplifies decision-making, but it still requires a person to decide how long each type of traffic is allowed to wait.
Alex Olson, acting head, U of T Centre for Analytics & AI Engineering
Next steps and timeline
The city will reveal more details in a broader $74-million congestion plan due in April. Contracts with technology vendors are still being decided. Officials say AI will not replace control centre staff, it will let them focus on complex corridors while autonomous systems manage routine, high-volume adjustments.
Council will consider the budget request as part of the city’s final spending plan. If approved, rollout will begin on selected corridors and at strategic intersections, with Finch West signal work moving ahead as an immediate priority.
AI promises faster responses to changing traffic patterns, but officials emphasise that human oversight and careful rule-setting remain essential to ensure the system performs fairly and reliably across Toronto’s varied neighbourhoods.
trafficartificial intelligencetransportationFinch West LRTcity budget
