AI in Transportation

A Response to the ITE Technical Brief

Earlier this month, I had the opportunity to speak at the ITE Canadian District Conference in Halifax (June 1-4, 2025), where I presented Leveraging AI in Transportation Engineering. During conversations with fellow attendees, I was introduced to a timely new ITE publication: State of Play 2024: Artificial Intelligence in Transportation (SOP2024), written by Venkat Nallamothu. It's a concise, informative read and highly recommended for transportation professionals navigating this evolving space.

Background

If you did not see my presentation at ITE in Halifax, you can download it here. The session explored why AI matters in our profession, what tasks it can support today, the importance of human judgment, emerging legal and intellectual property (IP) concerns, and the policy shifts likely to shape our work in the coming years.

The session included five case studies that tested AI, with both promising and problematic results. These use cases included:

• ITE Trip Generation Hotel Land Use Lookup

• OTM (Ontario Traffic Manual) Bicycle Signal Head Placement

• City of London (Ontario) Traffic Signal SOP Guidance

• OTM Rb-79R (i.e. No Right Turn On Red) Sign Image Generation

• OTM Sign Recognition from Uploaded Images

These weren’t hypothetical demos; they were purposefully grounded in real tasks transportation professionals might face, which aligned closely with SOP2024’s call for practical experimentation and sharing best practices across the profession.

To accompany the Leveraging AI in Transportation Engineering presentation, an expanded white paper was created that delves further into the transportation-related case studies and includes full AI discussion transcripts and my own survey results.

If you’d like the full copy of the white paper, please reach out to me: aiwhitepaper@jkts.ca

Survey of Transportation Professionals

Returning to ITE’s SOP2024 publication on AI, I found that much of the background information and survey responses revealed common insights with my survey results, namely:

• AI adoption is real but fragmented, often driven by individual initiative rather than organizational strategy;

• AI is most useful for early-stage, repetitive, or format-heavy tasks, but not (yet) for final analysis or decision-making;

• AI readiness hinges on support that includes training, policy, and validation frameworks; and

• Ethical, professional, and legal risks are real and must be addressed.

While not coordinated, the survey results in my white paper and ITE’s SOP2024 technical brief tell a similar story: AI is gaining traction among transportation professionals, but support systems haven’t caught up. Both surveys underscore the need for training, clearer policy, and human oversight. My findings also surfaced a distinct generational divide in interest.

These findings corroborate the two surveys anecdotally, illustrating that the data are on the same track. There is more to discover from transportation professionals, and it will be interesting to see how both AI technology and user attitudes evolve in the coming years.

SOP2024 left off with five considerations and next steps that the ITE is pursuing and will be the subject of the remainder of this post.

Education and Knowledge Transfer

SOP2024 discussed the development of clear definitions and a glossary of AI terms relevant to transportation professionals. I think this is a great idea and believe that they should be in plain language and defined as such, so that they relate to engineers and planners working in our profession.

During my presentation, one colleague remarked that while many of the over-40 crowd seemed more reserved (and often distracted with their phones), the under-40 attendees were leaning in to listen. This generational split over AI usage is real. The way we train new transportation professionals to work with AI, not around it, will be just as critical as the AI tools themselves.

Knowledge transfer through AI also means ensuring that “passing the torch” doesn’t inadvertently hard-code outdated assumptions, ideas, and thinking into future AI tools.

Evaluating and Validating AI Tools

The ITE is a respected organization in our profession, and its presence in identifying “deployment-ready” AI tools would be a similar stamp of approval and authority to that of the Trip Generation Manual. However, the ITE will have to walk a fine line in endorsing specific tools, given the potential legal and liability risks of AI tools and their outputs.

SOP2024 calls for frameworks to evaluate AI tools in transportation, which is a much-needed initiative. But the reality is that AI development is moving faster than traditional committee timelines. I'm not saying structured review is obsolete (far from it), but we should also ask whether AI itself might be part of how we rethink those very processes.

One of the clearest risks is not in using AI, but in using it without understanding when and how it fails.

Sharing Best Practices and Resources

This is where the ITE and similar organizations have a great opportunity to leverage their established platforms to gather, vet, and share the best AI knowledge, practices, and resources in our profession. SOP2024 recognizes this and expands on the idea, encouraging broader collaboration and more structured knowledge exchange.

The case studies I shared (mentioned above) were an exploratory process to understand the good, the bad, and the ugly of AI in transportation today. These kinds of real-world examples are critical to moving beyond hype and into practical, responsible use.

Webinar series, knowledge hubs, and peer-led discussions will need to cover not just the visible, field-deployed technologies (e.g. intersection sensors, incident detection, real-time insights, etc.), but also the soft skills that allow transportation professionals to add value (e.g. proposal and report writing, peer review and insights, data analysis, etc.) and, most importantly, the human touch.

In my presentation, I gave our AI companions the persona of an “EIT” – fast, capable, and eager, but not yet ready for a stamp. Human junior professionals will still need opportunities to “practice, practice, practice” with AI as a collaborator, not a substitute. We’re entering an exciting phase where sharing not just outputs but how we work with AI will define best practice in the years to come.

Policy and Standards

In my white paper, I highlighted how AI legislation is beginning to emerge, including Ontario’s Bill 194, which outlines requirements for AI use disclosure, human oversight, and risk management in public-sector services. Other jurisdictions across Canada and the United States are taking similar steps, signalling that regulatory expectations are no longer hypothetical.

SOP2024 rightly calls on ITE to engage in these conversations. As policies evolve, key issues will include how AI use is disclosed, how outputs are validated, and how risks related to privacy, bias, and accountability are managed. This isn’t just about compliance; it’s about ensuring that consultants use AI ethically, public agencies maintain trust, and the public remains protected.

ITE has a meaningful role to play here as a guide and can help set practical standards that reflect innovation and professional integrity, from helping define AI procurement disclosures to supporting members with validation frameworks for AI outputs used in project deliverables.

Broader Initiatives

Lastly, the SOP2024 publication highlights the importance of looking beyond transportation to accelerate AI readiness. The opportunity here is twofold:

• First, share what’s working (and not) from within our profession; and

• Second, borrow from industries that are further ahead in their AI maturity.

If ITE members can skip the trial-and-error stage by learning from industries, for example, logistics, healthcare, energy, or even municipal governance, then it's worth the effort. Cross-industry scans, shared case studies, and ongoing monitoring of open-source tools and emerging models will be essential. These learnings can help guide more confident, ethical, and technically sound applications in day-to-day transportation work.

We don’t need to build every AI tool from scratch, but we do need to build the practices and frameworks that make them effective and mitigate risk.

Conclusions

The conversation around AI in transportation is just beginning to emerge. The SOP2024 technical brief, like the conference presentation in Halifax, shows us a profession in transition: curious, cautious, and increasingly engaged.

There are both risks and momentum. I’m hopeful that we will continue to share what’s working, call out what isn’t, and build the skills, frameworks, and systems we need for AI to support (not replace) our professional practice.

Which side of the AI divide do you fall on - promoter, detractor, or fence-sitter? How have you personally used AI in your work as a transportation professional? I’d love to hear from you on this topic!