Planning With Artificial Intelligence

Today, artificial intelligence (AI) is altering the human experience, causing profound social, economic, and technical changes that touch all of us. The convergence of very large datasets and the capabilities to analyze them represent a potentially transformative period for the planning profession.

Now is the time for planners to examine the responsible deployment of AI to maximize its benefits and reduce its drawbacks, particularly in the face of its existing uncertainties. Having general knowledge of what AI is and how it operates, as well as understanding the most important practical and ethical aspects surrounding its use, is essential to implementing AI in ways that best serve planning practice.

AI holds the potential to automate repetitive, labor-intensive tasks in urban planning, allowing for more efficient decision-making in the planning process and freeing up planners' time to focus on higher-value work. However, not all facets of planning are suitable for AI applications. To identify opportunities for improvement, planners will need to assess their procedures against AI capabilities and determine the necessary steps to implement changes. As AI continues to advance, its use in urban planning is expected to expand, creating a need for awareness and knowledge among planning practitioners, researchers, and educators.

This PAS Report introduces planners to basic AI methods and concepts, and it provides current examples of how AI is already being used to augment planning activities. It seeks to provide the information planners need to begin recognizing how AI methods might fit planning tasks that they are working on. Planners may not be responsible for building AI solutions, but if they understand the general framework and the language of AI, they can better communicate with application developers and data scientists.

AI TOOLS IN PRACTICE

This PAS Report focuses on four primary application areas of AI: machine learning, artificial neural networks, natural language processing, and computer vision.

In machine learning (ML), machines (i.e., computers) detect and "learn" patterns, and then use this information as part of decision-making processes. ML is used to enable functions such as sentiment analysis, which examines text to determine the attitude that it conveys (e.g., positive, negative, or neutral); spam detection, in which algorithms learn to categorize email messages as valid or spam and filter out the latter; recommendation systems, which use a consumer's prior purchases, actions, or characteristics to anticipate additional products the consumer "may also like"; and predictive maintenance and anomaly detection, which predicts the likelihood that machinery will break down, determines maintenance scheduling, and monitors and identifies equipment problems in real time.

Artificial neural networks (NNs) are computerized systems based on how the brain processes information that can form connections between two items and learn to correlate them with one another to make predictions. NNs are used in speech recognition, which translates spoken words into text for virtual assistants and live captioning, and general pattern recognition, used extensively in fields such as biology, psychology, health, and marketing.

Natural language processing (NLP) comprises computer methods that analyze human language, text, or verbal communication to derive meaning. NLP systems analyze the ideas conveyed by words, and they can also correct grammar, convert recorded speech to text, and translate between languages. NLP combined with ML is used to operate AI virtual assistants, such as Apple's Siri and Amazon's Alexa. NLP also drives text analysis and content creation by systems such as large language models like OpenAI's ChatGPT.

Computer vision (CV) focuses on enabling computers to interpret and understand visual information, such as images or video. CV applications include raw image data processing, such as the conversion of satellite imagery to output maps, and detailed image analysis, used in facial recognition systems and other applications. CV is increasingly being used in conjunction with NNs in the transportation sector for systems that can classify and track vehicle types and roadway usage, monitor pavement and infrastructure conditions, and allow self-driving vehicles to "see" and respond to their surroundings.

ML, NNs, NLP, and CV can be combined to create AI-powered analysis tools that can help planners process large amounts of data, identify patterns and trends, and make better- informed decisions. These application areas are also being used to develop smart city technologies, which can automate and improve city services such as transportation, energy and water management, public health, safety, and security.

This PAS Report offers specific examples of how these AI applications are currently being used in the field, including the following:

• In Tuscaloosa, Alabama, CV is being used to assess neighborhood conditions throughout the city. Smart cameras with GPS-based image collection mounted on municipal garbage trucks collect image data from neighborhoods, which is processed using an ML model to generate blight scores. This has allowed for efficient identification of blighted areas and targeting of activities and resources for blight intervention and remediation planning.
• In Ector County, Texas, a school district used AI methods and spatial analysis to improve social equity and incorporate school districts as significant partners in community planning. By analyzing the socioeconomic characteristics of the district's physical environment, planners were able to visualize the district's standing and identify areas in need of improvement.
• In Long Beach, California, NLP is being used to capture and evaluate public input. More than 14 departments use the city's sentiment analysis platform, which collects data from social media, news sites, and city hotlines to analyze residents' conversations and categorize comments by topic and sentiment (positive, negative, or neutral). This approach helps quantify attention to specific issues and evaluate community perceptions. NLP is also being used in New Zealand to engage and communicate with Indigenous people in university and municipal settings.

The report also explores how researchers are using NNs to analyze transportation data to detect travel impacts and study street network design, disaster and evacuation route planning, health and safety, and social equity; analyze and model land use and land-use change; and inform urban design and development.

CHALLENGES TO AI ADOPTION

The adoption of AI in urban planning presents both challenges and opportunities. This PAS Report examines seven key challenges to AI adoption within the planning profession: fear and uncertainty, the need for new skills, changing data needs, unclear goals, transparency and explainability, bias, and ethical considerations. By proactively addressing these challenges, planners can harness the power of AI to improve planning processes, drive innovation, and ultimately create more sustainable and equitable urban environments.

The report highlights setting a clear vision, organizational goals, and performance measures as key components for effective AI implementation in urban planning, as well as the critical role of high-quality data and data management skills. It also emphasizes the importance of transparent and explainable AI systems for public understanding and acceptance, and it reiterates the vital need to address bias in AI systems and develop ethical standards for responsible AI use that center fairness, transparency, privacy, and accountability.

LOOKING AHEAD TO AN AI FUTURE

The use of AI is gaining momentum and offers a wide range of benefits for creating vibrant and sustainable communities. AI will not replace planners; instead, it will provide a useful set of tools for a subset of planning tasks and enhance planning practice by allowing planners to focus on the human components of planning.

How can planners begin to integrate AI into their planning work? This report suggests that planners consider five areas of strategic intervention in planning — community visioning; plan making; standards, policies, and incentives; development work; and public investments — as a starting framework to identify areas where AI can be applied. The report also recommends developing a strategic organizational plan for the adoption of AI, which should include a thorough analysis of the current state of the organization, identification of potential areas for AI adoption, and a clear vision for the implementation of AI, including addressing potential challenges and ethical considerations.

In late 2022, conversational chatbots with advanced language features gained significant attention. These chatbots, such as OpenAI's ChatGPT, use natural language input and can produce various output styles such as blog posts, essays, and recipes. As AI algorithms and ML techniques continue to advance, these may be some of the first AI tools widely used to support data-driven and efficient planning. The adoption of AI tools in urban planning will be incremental as skill development occurs through professional training and university programs.

AI is an emerging technology that is here to stay, and it will play a significant role in the future operation and planning of cities. Despite the challenges, AI applications have the potential to improve the understanding and development of more habitable and equitable urban spaces. As more and more data is being generated all around us, urban planners will increasingly use AI to analyze and process these data to detect patterns, make predictions, and have a better understanding of urban dynamics.

This PAS Report explores the substantial opportunities and challenges presented by AI technologies, with more being revealed each day. Responsible and ethical AI use can help planners in their work, enhance current planning procedures, increase efficiency, and allow planners to refocus their work on the human components of planning. In reading this report, planners will begin the process of increasing their knowledge and skills with appropriate AI methods to enhance planning practice in the service of creating more livable, resilient, and sustainable communities.