Philly’s poverty rate won’t budge. Could artificial intelligence help?

Between 2006 and 2016, Philly’s poverty rate moved a total of .6 percent, and in the wrong direction.

While over that decade the rate ranged from a low of 23.8 percent and as high as 28.4 percent. All in all, over the last 10 years, the rate’s been flat. At the systematic level, have we really moved the needle in support of this population?

As much as Philadelphia’s impact ecosystem of nonprofits, B corps, government agencies, social entrepreneurs, religious institutions, philanthropists, foundations and impact investors do every damn day, as great as some of our outcomes are — collectively, are we just fingers in a failing dam?

For everything Philly’s impact ecosystem accomplishes, for the many people who are helped, is it enough to fight the pull and churn that drags people into poverty every day? But are we really an ecosystem or a field of silos, each with its own priorities, data systems, alliances and yes, grudges, forced to compete against each other to help who we can, how we can?

We have to change what we do and how we do it before we normalize poverty in Philadelphia.

We have to change what we do and how we do it before we normalize poverty in Philadelphia. Based on the poverty rate, there are roughly 400,000 residents who live below the poverty line. Understanding how to fight it should not be considered a programmatic issue, but at the individual level.

With available technology and the data that gets collected on everyone, it should not be hard for us to create to individual action plans, tailored to the needs of each single individual person living under the poverty line.

If Alexa and Siri can anticipate our needs, there no reason we couldn’t build a AI Case Worker, always available and ready to help, to make suggestions on everything from after-school programming to getting your criminal history expunged. Some very basic AIs can fight parking tickets for you; another is already working to help people immigrate to the U.S.

Philadelphia’s Data Management Office, is a central component of the Office of Deputy Managing Director of Health and Human Services that operates and maintains the CARES integrated data system, which receives information from 11 unique source databases. Philly uses this system to do make policy decisions, to understand vulnerable populations and how best to serve them.

Many cities and counties have integrated data systems, partly thanks to the advocacy of the Actionable Intelligence for Social Policy, an initiative run out of the University of Pennsylvania. These systems are mostly used at the macro level and not at the individual level, although there are some cities experimenting in using data for more direct action.

This is new territory, and it carries with it all sorts of bias issues that have to be figured out, but this kind of work will be universal soon.

Allegheny County in Pennsylvania, for instance, combined eight databases to create predictive risk modeling to identify potential child abuse based on calls to local child welfare dispatchers who must vet thousands of calls a year. The software helps the screeners sort though massive amounts of data in seconds to produce a score that helps the them decide who to visit.

And Chicago created an algorithm that is predicting who will be involved in gun violence, either as a victim or perpetrator, creating Strategic Subject List, where those with high scores can be identified with specific interventions. This is new territory, and it carries with it all sorts of bias issues that have to be figured out, but this kind of work will be universal in the next five to 10 years. [Editor’s note: This reminds of the City of Philadelphia’s grappling with its own risk assessment tool.]

But what can we do right now? We can still break down those 400,000 Philadelphians into manageable sizes: those under 18 and over 65, those who are disabled and receiving SSI benefits, the unemployed, those in jail. Each category may have some crossover, but real change can start looking promising when we visualize ways to support these subsets.

For example, the largest subset, the youth, could be further broken down by age. It should not be hard to support every child with age-specific interventions, making sure they are in a daycare or an after-school program; getting mentoring and tutoring; getting a summer job; graduating from high school; applying for scholarships; or getting into a technical, college or apprentice program. Many already have access to these types of programs, but it is not being done in a holistic, comprehensive way.

A birth-to-career pipeline could be built out of the existing elements of Philadelphia’s impact ecosystem. The data that the city is already collecting needs to be better shared with the nonprofit sector, and the nonprofit sector should be feeding good data into these systems so that we can create comprehensive and customized individual integrations that would create real and lasting impact to finally, truly move the needle.