It’s time for a post on artificial intelligence (AI) and affordable housing. The topic is interesting but AI comes with a great deal of hype. First of all, based on my experience with OpenAI’s free version of ChatGPT, that the world has to fear being destroyed by that AI. I know. It’s free. And there are more sophisticated versions behind pay walls and being developed for proprietary use or who knows what else. As with any data tool, however, the garbage in, garbage out principle still applies. The AI is only as good as what we ask it and what it is allowed with its programing to answer back. For this post, I’m using three sources, one human and two AI. One AI source is from a blog at the World Bank written by AI, and the other is my own free ChatGPT AI.

The human is Christian U. Haas, CEO of PTV Group. Haas has more than 20 years of international management experience in the software and technology industry. As CEO of Umovity, he leads the merged PTV Mobility and Econolite brands to provide global Intelligent Transportation System (ITS) products and software technologies. The field of ITS – using technology to improve things like traffic flow and freight mobility – has been around for years. Today, of course, there is a move to incorporate AI into setting up systems to make transportation more efficient. But let’s start with the World Bank Blog.

Simon Walley, Lead Financial Sector Specialist, Finance, Competitiveness & Innovation Global Practice and Olivier Vidal, Financial Sector Specialist with the Bank and Olivia Nielsen, Principal at Miyamoto International asked ChaptGPT (it’s not clear which version) to:

“Write an inspiring blog of (approximately 900 words) on how AI can help promote access to housing, with a focus on emerging markets. Provide specific country examples. Highlight how AI can enable informal households to access finance. Highlight how AI can help decrease housing costs. Showcase how AI can help make housing green and resilient in the face of climate change.”

The result is an earnest effort using citations of United Nation’s data and sort of repeating various other statements that were clearly mined from various other written sources on the internet. Most of ChatGPT’s answers or efforts usually sound like a well written freshman college essay, regurgitating lecture notes, referring to the texts in the class, but not offering much compelling insight. For example, this paragraph.

“By analyzing data on housing demand and supply, as well as data on the social and economic characteristics of households, AI algorithms can identify areas where there is a high demand for housing. For example, in India, AI algorithms can analyze data on population growth, urbanization, and migration, to identify areas where housing demand is likely to increase in the future”

Sure. But so can really smart people. In fact, that’s exactly what everyone in the real estate world tries to do every day: predict where people will go, how much money they will have to spend, and what kind of housing they’ll want and need. It also made this more interesting observation.

“AI can also play a role in enabling informal households to access finance for housing. Machine learning algorithms can be used to analyze financial data and assess the creditworthiness of households, which can help to identify those who are most in need of financial assistance. For example, in Latin America, AI-based microfinance platforms can be used to provide loans to informal households who would otherwise have difficulty accessing traditional banking services. This can include providing access to microfinance loans or other forms of financial assistance to help households access housing.”

This is important not just in Latin America, but in the United States as well. Traditional financing has failed to create any significant changes in patters of homeownership; the gap between Black and White families is still large and hasn’t shifted since the signing of the Fair Housing Act. But ChatGPT picked this up from ongoing efforts at the World Bank. Here, AI is just pointing at what’s already being done, not coming up with a new idea.

So, I had a question for ChatGPT as well as for Haas.

“Do you think that Artificial Intelligence, properly applied to smart cities, could provide a useful tool for both governments and consumers when planning and thinking about housing? Could Technology/AI reduce the marginal disutility of moving, for example?”

Here’s how it answered in part about reducing the disutility associated with moving.

“AI can play a role in reducing the challenges of relocation by optimizing urban planning and enhancing accessibility. AI-powered tools can help identify areas with desirable amenities, efficient transportation options, and improved infrastructure, making potential new locations more appealing. Predictive modeling can assist in creating urban environments that cater to individuals’ preferences and needs, mitigating the disutility associated with moving. Additionally, AI-driven data analysis can aid in developing strategies to minimize disruptions and facilitate a smoother transition for individuals and families, ultimately encouraging mobility and contributing to more livable cities.”

Here, in part, is Haas’ answer.

“Data-driven software also excels at identifying areas with a shortage of affordable housing, a poor local supply, or weak accessibility. For instance, at PTV we leverage data analytics and machine learning through one of our SaaS products called PTV Access. The automized and standardized process enables our customers to gain valuable insights into the quality of transport infrastructure. This encompasses evaluating the accessibility of various city areas through different modes of transportation, including public transit, cycling, and walking. It also assesses the ease of reaching key points of interest such as schools, hospitals, or shopping centers. In Germany, for example, we developed a web-based digital model for assessing service provision in rural areas for one of the Federal States. It enables local governments to assess the accessibility of essential services and identify areas where improvements or investments in housing and mobility infrastructure might be necessary.”

What’s interesting here is that Haas sees the potential of AI being part of helping people make decisions about housing and mobility and both instances of ChatGPT, the World Bank’s and mine, made the same connection. And it makes sense. Better understanding of people’s transportation needs is strongly linked to housing choices and needs. Years ago, I wrote about how housing policy is transportation policy, and vice versa; housing location is critical for utility and affordability. People make choices about where they live based an array of considerations, not just price.

This little experiment convinces me that we’re better off working and trusting living breathing experts like Hass over AI that currently seems to just spit out very nice college essays that would earn a B+. Getting government to develop better and more transparent data tools using technology like Haas’ could help them make better decisions and improve affordability. But AI is not going to save us or destroy us any time soon; only human policy makers can do that.