Ever wondered what Artificial Intelligence might do for urban planning? Look no further than “A Digital Future” (Thurs, 1 September) – NZPI’s Special Virtual Event.
Starring in this online think-fest (amongst a galaxy of others) is Tom Sanchez, Professor of Urban Affairs & Planning at Virginia Tech (http://tomwsanchez.com). We spoke to him recently, following the publication of his paper on The Prospects of Artificial Intelligence in Urban Planning (Sanchez et al) in the International Journal of Urban Science.
The paper delivered results of a survey – “the first of its kind,” says Tom – into the knowledge and uses of AI by planners in the USA, with an international literature review adding to the picture. It concluded there was a “tepid” adoption and endorsement of AI amongst planners currently, although most also agreed AI would impact significantly in future.
So how could AI possibly help planners? And what’s available now that planners are not using?
“In theory, AI could give us better evidenced based solutions,” says Sanchez. “The outcomes and benefits to the public are better with evidence-based decision making. So I see AI as an opportunity for us to go through decision making processes and deconstruct them, and ask: ‘‘what did the process involve?’ ‘what else?’ ‘who were the stakeholders and what do we know about them?’ ‘how was their input used?’.”
“If we’re trying to better understand the real humans that sit behind our processes, we’ll be able to use AI to understand the biases that we’ve previously introduced, or understand where there have been breakdowns, or augment our processes in some way.”
Better and more data, he says, brings transparency and explicability to decision making – a kind of rational foundation to policy and plan-making.
But what kind of data, and where do you get it from?
“A big part of getting planners to use AI, is preparing them to better understand the data science on what’s going on in our cities,” he says. “There are new sensors, data collection, video surveillance. This data is not just used for identification or recognition (i.e., crime prevention). It’s used for counting people, analysing behaviour, looking at flows. As a result, we’ve found computer scientists and engineers taking the lead in urban computing. Many planners lack the technical background, so they can’t engage in the conversation. That’s one of the things I’m interested in. Planners need to be better prepared, more knowledgeable about these disciplines, and around data, data collection, the analysis, how it’s presented and might be used in policy.”
The phones people have in their pockets are mini collectors of data and behavioural information, he argues. Many commercial organisations already use this data legally in ways that promote their services, products or retailing, cohesively analysing customer behaviour and deploying strategies that interface with that behaviour, including through design of premises.
“The devices in our pockets collect information about where we are, what we do – every minute,” he says. “They know our contacts, what we search for, what we purchased. They monitor us – until we turn that monitoring off.”
Increasingly, too, cars, roads and transport are loaded with censors and data gathering tools, which could also inform planning decisions and processes.
“Transport is basically a data collection device in its own right,” he says.
But what should planners do with this information? Do they understand its value?
Tom likens the stage planners are at with AI and ‘big data’ as similar to that when GIS systems were first introduced.
“Originally,” he says, “GIS was a better way to draw maps, rather than by hand. Little thought was given to how this was a tool which could be used in other ways. Now, GIS has grown enterprise wide and it’s much better integrated.”
“With AI we’re at the same kind of early stages of GIS.”
Tom argues that AI can provide information without threatening privacy or compromising confidentiality.
“If planners had an understanding of people’s behaviour, their preferences, what they like, what they don’t like, how they need to conduct their business – we could plan to meet their needs,” he says, adding that there is certainly a line not to cross. “We as planners need to think: how much do we want to know? And how much do we want to leave to chance?”
Another aspect of providing big data to planners is understanding how to integrate with their processes.
“The whole idea about AI in planning for me is: how can we do a better job? How can we better meet the needs and build better quality communities? So it’s starting with the planner to say: if you had more information, what would it look like? If you had more expert input, who would be involved? An example I use is when someone who has been a planner for 20 or 30 years and leaves, they take all their knowledge with them. Could AI capture their knowledge and use it for the next generation of planners to those mistakes and errors we’ve made before are not made again? Can we move our processes forward?”
“We must be ready to try and do the next right thing. We can’t simply be comfortable with the way things are done now, and not consider doing it differently.”
