This week’s episodes of the What is the future for cities? podcast have given us much to consider about artificial intelligence in city planning. Episode 367 presented a debate on the 2025 paper by Haishan Xia and colleagues, titled “The fundamental issues and development trends of AI-driven transformations in urban transit and urban space“. This discussion featured an exploration the pros and cons of AI in urban systems. Episode 368 followed with an interview featuring Josh Rands, co-founder and CEO of TerraCity, who shared practical insights on using AI for transportation and land use. Together, these episodes reveal AI’s transformative potential and its challenges. Here are five key lessons drawn from the discussions, connecting the debate’s theoretical insights with Rands’ applied approaches.

Lesson 1: AI can uncover hidden patterns in urban complexity, but transparency is essential

AI shifts cities from static to self-evolving systems, analysing non-linear relationships like traffic, housing, and behaviour that traditional tools miss. Tools like gradient boosting and Shapley Additive explanations (SHAP) clarify variable impacts, countering AI’s “black box” issue. For instance, AI can pinpoint where rail stations boost land values, enabling precise investments to curb sprawl. Hangzhou’s City Brain, slashing congestion rankings, and Pittsburgh’s Surtrac, reducing pedestrian wait times by 25%, show AI’s power. Rands’ TerraCity models integrate inputs like weather to predict outcomes, revealing hidden patterns. Transparency ensures planners trust AI’s logic, making it essential for accountability. This precision aligns urban systems with human needs, but without clear decision pathways, AI risks losing credibility.

Lesson 2: Historical data biases in AI could worsen urban inequalities if not addressed

AI’s reliance on historical data risks embedding societal flaws, worsening inequalities. Optimised rail systems may improve access but trigger gentrification, displacing transit-reliant residents. Digital tools like contactless ticketing exclude those with low tech literacy, skewing models towards affluent users and misallocating resources. Training AI models also emits significant CO2, questioning short-term gains against long-term benefits. Rands’ TerraCity counters this by using comprehensive data to assess demographic impacts, simulating how bus routes affect neighbourhoods to adjust for fairness. Careful data design is crucial to prevent AI from reinforcing economic biases over equity. Quantifying effects alone isn’t enough; addressing biases requires proactive intervention to ensure urban transformations don’t deepen divides.

Lesson 3: Human-AI collaboration is key to balanced urban decision-making

A human-AI equilibrium integrates ethical oversight into AI’s analytical power. Simulations, like testing rent controls, anticipate imbalances before implementation, fostering urban harmony. However, costly infrastructure like digital twins and weak privacy regulations pose challenges, especially for developing cities. Rands uses AI to quantify emissions or land use shifts, relying on planners for priorities. His transit-oriented designs, like 15-minute cities, make essentials walkable, reducing car reliance. His nonprofit, Litter Cleanup, shows tech empowering community action. AI processes complex data, but human ethics ensure fairness. This partnership moves from reactive to proactive planning, vital for managing urban complexities as climate and population pressures grow.

Lesson 4: Innovation in urban planning requires risk-taking and a focus on long-term impacts

Innovation means risk-taking in cautious public sectors, as seen in programs supporting Rands’ AI-driven rail projects. Technology spans AI analytics, micro-mobility like scooters, and land use strategies like transit-oriented development. AI models non-linear dynamics—climate, population—that older methods can’t handle. Short-term policies or agendas block progress, but quantifying transit’s broader benefits, like emissions cuts, shifts focus to enduring impacts. Rands’ “true cost accounting” evaluates full effects, guiding fair decisions. His European-inspired vision contrasts US car dependency, promoting efficient systems. Progress requires bold risks, ensuring cities balance vitality with environmental goals for lasting change.

Lesson 5: Quantifying ripple effects enhances planning

AI quantifies secondary and tertiary impacts, unlocking better urban planning. Rands’ TerraCity models use ridership and traffic data to predict outcomes of new bus lines or developments, capturing effects like CO2 reductions or resource shifts beyond farebox revenue. For example, AI estimates emissions saved by fewer cars. The debate noted similar precision, like rail stations’ impact on land values, supporting targeted investments. Projections of 31.73 million tons of CO2 reductions from optimised traffic highlight the scale. These models integrate diverse inputs, simulating scenarios to inform decisions. Initial AI training costs are investments, offset by preventing gridlock. This data-driven approach transforms cities into adaptive entities, crucial for managing complexity.

Both episodes suggest urban futures depend on collective effort. Rands encourages curiosity and local contributions for brighter outcomes. The debate warns against tech outpacing ethics; Rands’ work offers tools to align them. Ultimately, this lesson reminds us that AI’s potential shines when rooted in community needs and comprehensive data, fostering cities that adapt and thrive for all.

These five lessons from episodes 367 and 368 provide a roadmap for navigating AI in urban contexts. From transparency and bias mitigation to collaborative models and inclusive innovation, the podcast illuminates paths forward.

As cities face growing pressures, these insights encourage thoughtful integration of technology with human values.

---

Next week we are investigating transportation as the movement of people and goods, instead of vehicles, with Josh Rands!

---

Share your thoughts – I’m at wtf4cities@gmail.com or @WTF4Cities on Twitter/X.

Subscribe to the What is The Future for Cities? podcast for more insights, and let’s keep exploring what’s next for our cities.