Demystifying Artificial Intelligence in Planning

Does the phrase 'artificial intelligence' evoke for you Hollywood images of dystopian futures filled with menacing cyborgs, sentient operating systems, and ubiquitous surveillance? (Oh my!) If so, you're not alone. Nearly half of the APA members who participated in our recent survey on the subject indicated that they had neither interest nor experience in AI.

As researchers studying prospective applications of AI in urban planning, we recognize the potential for algorithmic bias, invasions of privacy, and other negative consequences. However, we believe the best protection for citizens is an informed urban planning profession able to understand and evaluate the risks and benefits of AI applications for their communities.

AI in Planning: Potential and Equity

Most people encounter AI every day. AI populates your social media feeds, provides recommendations, computes the fastest routes to your destinations, screens your spam calls and emails, and protects your information from hackers behind the scenes. It's no surprise, then, that APA Foresight staff named AI as one of the "7 Trends Knocking at the Planning Office Door" last summer and listed it for the second year in a row as a trend we need to prepare for in the 2022 Trend Report for Planners.

In the private sector, companies deploy AI to save customers time and money, help them navigate complex processes, and protect them from bad actors.

Could urban planners use these same technologies to benefit our citizens? If so, how do we ensure they lead to more equitable outcomes and don't entrench discriminatory systems?

These are the questions motivating our research as part of a National Science Foundation-funded collaboration between Virginia Tech, APA, and Arlington County, Virginia. Our goals are to:

1. Identify the urban planning and decision-making processes with the greatest potential to utilize AI methods.
2. Develop and test pilot AI-related planning applications.
3. Pay special attention to the equitable deployment of these AI methods.

As the painful legacy of redlining and urban renewal reminds us, our traditional tools and techniques are equally capable of perpetuating inequality and enabling abuses of power.

It is our professional duty to ensure that new tools and techniques correct past and present inequalities rather than perpetuate them.

If we as a profession fail to take ownership of these questions when it comes to AI, then technology vendors will determine them for us from a different and possibly incompatible code of professional ethics.

Planners' AI Perspectives: Interest, Impact, Diversity

As a first step, we need to understand the current state of AI in urban planning. In the fall, we invited APA members to answer questions about AI including their familiarity with the topic, their expectations for its impact on cities and planners, and their ideas for its most and least productive uses within the field.

We received 396 responses from a diverse group of planners, geographically, demographically, and about educational backgrounds. Consistent with the demographics of APA members, participants more often identified themselves as male (63 percent) and as white (76 percent). Respondents hailed from every part of the country (and a few other countries as well), worked in both public (65 percent) and private organizations, and possessed all levels of experience: 18 percent of the respondents had 0-5 years, while 27 percent had 26-plus years. In this set of dashboards, we present some of our most interesting and important findings.

Survey Findings

First off, as we noted earlier, nearly half (47 percent) of planners who responded to the survey rated their 'level of interest or experience with artificial intelligence related to planning' as zero on a zero to five scale, with zero meaning no experience or interest and five meaning expert-level experience or interest.

All told, more than 80 percent of respondents rated themselves zero, one, or two on the AI interest/experience scale, and just 8 percent rated themselves a five. This disconnect is especially striking given that more than 70 percent of planners indicated that they believe AI will have an impact on urban planning, placing its significance at a three or higher where five denotes 'dramatic' impacts on the profession.

Beyond personal interest and experience, the varied perspectives about AI among APA members were underscored by planners' responses to a survey question asking them to define AI, which included a broad range of ideas. The word cloud below summarizes the most common words used in planners' definitions.

Common words planners used in survey responses to define AI.

Planners Explore AI's Vast Potential

Most respondents explained the concept by considering the relationship between computers, data, and humans. Specifically, they discussed whether and how computers can achieve and even surpass human-level thought and decision-making capabilities through complex analytics, algorithms, software programs, and the like.

While we know that many of those surveyed don't yet have much experience with AI, it's clear from their definitions that planners have already begun to form conceptions of what AI is and can do — some general ("using technology to hasten data collection and decision making"), others narrow (autonomous vehicles, code/permit review, AR/VR), and others deeply skeptical ("robot overlords," "privacy invasion").

Amidst this new, uncertain AI frontier, some planners are already implementing AI into their workflows, and they shared their experiences in the survey. The plurality of current applications come in the subfield of transportation planning, which may not surprise you; this was the field that our survey respondents flagged as the area with the highest potential for AI. Applications in this arena include: analyzing driver behavior from GPS trajectories, predicting areas of high or low safety for pedestrians and bicyclists from anonymized trips, traffic monitoring, fleet management, and routing of transit services.

Unexpected AI Applications

Respondents also described some perhaps more unexpected applications for AI in planning:

• Deploying chatbots on a planning agency's website to answer frequently asked questions.
• Using machine learning to parse public comments into thematic categories and run sentiment analysis.
• Leveraging algorithms to augment the decision-making process around siting new land uses like accessory dwelling units and large infrastructure projects (in addition to broader adoption of e-permitting and zoning management software).
• Using advanced visualization techniques, often enabled through drone and LiDAR collection, to model the possibilities for redevelopment.

Even in this relatively small sample of planners implementing AI, the applications ranged widely and still only skimmed the surface of the possibilities to come.

Prepare for AI's Transformative Potential

Artificial intelligence is quickly becoming ubiquitous in our lives, and planners have the chance to shape its adoption and use in cities in a way that promotes an equitable vision for the future. Today, though, as APA's 2022 Trend Report for Planners highlighted, "Technological shifts and innovations are still the trend category with the lowest preparedness among planners."

With that in mind, it is more consequential than ever for planners to begin to engage meaningfully with emerging technologies and grapple with both the prime opportunities to bring AI into our work as well as AI applications that require more skepticism and caution.

APA is collaborating with experts in the field to help you prepare for AI and its potential implications for planning. Stay tuned for more!

The survey reported here was supported through grants from the National Science Foundation's Smart and Connected Communities Program (Award #2125259) and the Virginia Tech, Institute for Society, Culture and Environment (ISCE) Scholars Program.

Top Image: iStock/Getty Images Plus - Ilya Lukichev