The American Planning Association (APA) in 2021 established the “AI in Planning” Foresight Community which, over ten months, surveyed and interrogated and analyzed and contemplated/deliberated upon the unfolding relationship between advances in artificial intelligence (AI) such as machine learning and the planning profession (this report tells you all you need to know for now - the technical details, big-picture developments, and ethical debates/dilemmas). This blog covers the core conversations and contributes my views on AI in planning education from the perspectives of both teacher and learner.
The AI project in planning seeks to harness the computer’s overwhelming power in gathering and processing data input, including that on its own performance, extracting patterns, and performing calculations through digital replication of the planning logic. The prevailing understanding and consensus appear to be that 1) the technology will out-consume and out-power any resistance in its redefining the field much in the same nature as, only even more comprehensively than, geographical information systems (GIS) a few decades ago, and 2) its responsible use requires that planners learn its inner workings and, when applicable or appropriate, utilize its capabilities to make decisions and model projections for human settlements and communicate its operations to the affected communities. The creation by APA of its team of experts reflects the recognition that if planners are unprepared, they could either miss out on all the answers and solutions or misuse its output. Right now, the outlook is that technological advancement, largely in the private sector, will vastly outpace its adaptation for policy and planning in the public sector and scrutiny of its capability to do so effectively by universities and nonprofits.
The current disconnects hampering the effective use of AI by planners across all domains as identified by the academic experts through a combination of meta studies and case studies can be pinpointed at these particular loci of “dissonance”:
1) between any of a planner’s desire/will (what problem am I trying to solve), comfort/capacity (how well can I use the tool at hand), and awareness/comprehension (how appropriate is the tool to the purpose, and what is its limitation): until planners can unequivocally define exactly the procedures, goals, and core values motivating all decisions and choices in their scope of practice, AI will always remain an unstable concoction of uncertainty and powerful potential for mishap/mischief.
2) between private sector priorities, which shape the technology’s development and deployment, and public sector imperatives, which form the bottom line of re-distributive justice and co-existential compact.
3) between the respective timescales of innovation, financing/entrepreneurship institutional building and capital/labor routing, and knowledge production on its methodological possibilities and ethical implications across the various spaces or sectors – academic, nonprofit, governmental, corporate, tribal etc.
4) between the respective cognitive spaces of planners and their communities – be they settlements, workspaces, classrooms, or legislatives.
In essence, moving forward requires bridging the perception-understanding gaps around any social spaces where AI is relevant or a potentially advisable/preferable direction and taking assessment of what we know and don’t know about what we will and won’t do.
I’ve delighted in reading the cases of AI’s pilot for solving long-standing problems over humane and just resource allocation. I await large-scale, replicable studies of its effectiveness as a governance tool in all spheres of social existence. And we are all part of a nascent experiment with a hitherto unknown set of circumstances – namely, an epistemological transformation superimposed on various inherited structural dysfunctions…
- one of which being deficiencies in our teaching and instruction of the fundamentals of math and computing and philosophy and reasoning. That our entire education system in the country and parts beyond requires an overhaul is besides the point, but the only limit to perfecting planning educational programs’ purpose for preparing AI-knowledgeable prospective planners is lack of political will to overcome/disrupt institutional rigidity or inertia. What I’m trying to say, we have the individual and institutional capacity in our existing systems for producing things – knowledge included – to implement the leading recommendations, just that outdated thinking needs to move out of the way.
As a new teacher of urban planning, I have been fiercely outspoken against the use of AI in any kind of instructional settings where learning – as in gaining true comprehension – is the primary goal. A machine designed by humans for computing – in the broadest possible sense of the term – human-generated data (even nature measurements are only meaningful to the extent humans deem it worthwhile to measure) can confirm the established conceptions or replicate/augment human actions/calculations or justify chosen recourses or predict forthcoming occurrences based on known likelihoods - that it can do so is meaningless or irrelevant for gaining deeper awareness of our worldly and outer-worldly relations.
In other words, it has no place in the learning process other than summarizing a ridiculous amount of data that no known biological organism can consciously do according to the metrics we have agreed on. For this reason, I find entertaining any notion of incorporating AI in the classroom of planning education utterly risky as human minds, young and old, are far too impressionable, and for good evolutionary reason, to be trusted with ex-poste knowledge reconstruction. (Besides, making predictions based on statistical/probabilistic models and then retracing the logic in the models makes learning like chasing one's tail, only much less fun). My advice for students at all stages of life is: don’t let AI deprive you of the opportunity to make the connection yourself; you can always ask it afterwards.
The implication from that last statement is that there’s no theoretical gap between a thinking thing made up of metallic or organic matter or whatever combination of substances. That our machines will produce ever closer approximations and eventually replicate human intuition and ingenuity and imagination is not a question but an inevitability. Good news is that we are always moving toward convergence or some sort of new equilibrium following any kind of disruptive cognitive/technological advances; the challenge is minimizing the damage on the way -- that our experimentation with various AI projects don’t cause unnecessary harm or suffering. Students who mistake AI for an aid to learning beyond simple fact-finding and pattern-summarizing simply have not noticed the delight of true learning. The goal for us educators in the classroom is to recall, reproduce, and reinforce those fleeting joyful moments of epiphany and revelation – like that one when you first discovered negative numbers as a child, of learning through love and human connection, and thereby put AI in its rightful place of serving simply as an efficient, aided self-correcting gyroscopic calculator, to be used only by those properly trained.
Googling “AI planning.org” will return all the current AI endeavors in urban planning – the first few results being the newest publications on the current state of knowledge and application and deliberation, each providing links and citations to useful resources.
Comments