Scott Jensen is the chief executive officer and vice president of external affairs for Research Improving People's Lives (RIPL), a nonprofit that works with governments to help them use data, science, and technology to improve policy and people’s well-being. RIPL partners with policymakers to solve pressing social challenges and build their own capacity to innovate and measure success going forward. Jensen served as director of Rhode Island‘s Department of Labor and Training from 2015 to 2021 under Governor Gina Raimondo.
RIPL’s main initiative is the Data for Opportunity in Occupational Reskilling Solution (DOORS). DOORS uses machine learning (ML), artificial intelligence (AI), and data from state agencies to identify promising employment and training opportunities for jobseekers looking to earn more or gain new skills. DOORS has launched in Rhode Island and Hawaii and is under development in Colorado, New Jersey, and Washington.This interview has been edited for length and clarity.
Andrew Boardman: DOORS is an innovative program helping jobseekers match with employers and training opportunities. You liken it to “Netflix for jobs” and “Tinder for jobs.” Could you briefly explain how it works?
Scott Jensen: DOORS is a suite of technological applications that helps people reskill and find opportunity more efficiently. What does that mean in practice? It means there is an AI- and ML-driven, worker-facing search engine that you interact with, and based on information about you, it helps you find a new job by showing you job openings, training opportunities, and new career directions.
The “Netflix for jobs” piece is substantively important in a certain way. Yes, it’s kind of pithy and gives people the right image of the program. But most importantly, the interface is a display that that can say: because of your background, these things might be interesting to you, here are job openings that are like that.
AB: How does the system find good matches for jobseekers? What’s an example of an occupational switch or industry-to-industry move the data show has promise for a worker looking to advance into a higher-paying job but who currently works in a lower-wage position or is unemployed?
SJ: We look at unemployment insurance wage records, which most employers in the country submit quarterly. These records show how much individual employees are making. We’ll take 10 years of data from a state and look for people who switched industries. As you can imagine in a dataset that big, many people are switching industries. So we isolate what the causal effects of these industry switches are for the people who make them.
In Hawaii, we found about 33,000 industry switches that had enough data around them to make a causal determination. About a third of them were bad: “Boy, don’t switch from industry X to industry Y because if you do, you’re going to make less money.” About a third of them were without a lot of consequence one way another. But about a third of them were really good.
That information is good to know if you’re contemplating an industry switch. Because you know what industry you’re in, but now we can give you a forward-facing, data-driven, objective view of what’s going on in your labor market. It doesn’t require any prior knowledge or connections on your part. You don’t need to hear about a job from your cousin. You don’t need to know somebody who made such a switch to at least get the idea if the computer platform has surfaced it for you.
DOORS could also surface jobs that you didn’t know existed. Like, “Wow, I didn’t know that there were jobs like process technologist, which doesn’t need a college degree and it’s an $80,000-a-year job.” And if you’re detail oriented, you might want to do it. There are a whole bunch of opportunities we can quickly elevate through DOORS so people can see them.
AB: What are your metrics for success with DOORS? How will you know when the program has achieved some of its goals in a state, and what results are you seeing so far?
SJ: We launched DOORS in Rhode Island about a year ago. And Hawaii launched on March 31. We have about 5,000 users, so it’s a little bit early. But we just finished our live dashboard, where we’re going to be tracking every state that goes up. I would say after a year, we will have enough data to see what’s going on.
What we’re looking for is a return-on-investment [ROI] of employment transitions people make using DOORS. States that have launched DOORS know who uses it, know their Social Security number, and can locate them in the wage data. And we’ll be able to analyze the impact of a job move and see what the ROI is for folks when they use the DOORS product. We also want to watch, to the extent possible, the demographics of the people who are using it.
AB: What are some of the big challenges you’ve faced in setting up innovative programs like DOORS?
SJ: Over the last 10 years in the private sector, we have started to see applications of technology that are genuinely useful. You can buy a book on Amazon in, like, four clicks, and it shows up at your house, because it already knows your credit card number. I’m not a big fan of Domino’s pizza, but I have ordered Domino’s pizza just to watch the pizza tracker show up at my house. And that’s genuinely helpful.
That kind of thing can happen in the public sector as well. What’s holding it back is an appreciation that you can safely share data inside of government in a way that is controlled and that the people who are responsible for the agencies have a sense of what’s going on and how data is being shared.
The final thing is that the procurement structures inside of government make innovation difficult. It’s hard to do agile development when the process can bog you down and costs can escalate. These things make the innovation we’ve seen in the private sector harder to do in the public sector. But it can be done. And we have to do it.
AB: I saw that RIPL is also working on an employer-side portal for DOORS. How does that work?
SJ: The employer side of the DOORS platform is called Ready Hire. This is the “Tinder for jobs” part. If an employer has a job opening, we use AI to take that employer’s job description and the hiring manager’s understanding of the job, translate that into a skills ontology, and take out the degree part. Then we do the opposite with folks currently receiving unemployment insurance.
We then show a list of anonymous candidates to this employer and say, “Here’s a list of 25 people. Are you interested in talking to this person? Because they seem to be a good match.” If the employer says yes, we’ll send a text message or an email to the person. The person can then see company X thinks you’re a good match for a job. That job makes $40,000 a year. Are you interested in talking to that company, yes or no? And if the answer is yes, it’s like swiping. The contact information for the person will show up.
Now, the AI will never know exactly when someone is a match. All the things we build are predicated on good old-fashioned human judgement in the labor market. But we can make it quicker and more efficient by using technology well.
There aren’t any perfect silver bullets, but 10 percent, 20 percent, 30 percent more efficient in a big labor market is a good thing for everybody.
AB: More employers are exploring skills-based hiring to expand their talent pools and become more inclusive. The idea is that by emphasizing skills and capabilities over credentials, employers can unlock advancement opportunities for workers who don’t hold college degrees. Do you see DOORS as playing a part in facilitating this transition?
SJ: There’s a lot of talk about credentials and skills-based hiring. And when you start to think about that, the problem is that everybody knows what a college degree means, but that’s not always the case for skills and credentials. Saying “I have a credential” doesn’t help much because how does an employer know that credential is any good?
RIPL is attacking this a couple of ways. First, you need a computer platform where employers and potential employees are near each other. Our “Tinder for jobs” system is a good example: people are searching, companies are searching. We’re working on essentially a translation service for how people are describing their jobs. It’s using AI and machine learning to mine text and using topic modeling so we can translate the language an employer is describing their job with into a common skills ontology. And then we do vice versa, in terms of what forms of credentials and badging and experience someone has, so we get to this common language.
Second, we’re working in partnership with the Credential Engine and other partners in Rhode Island to improve the DOORS product to really do a great job in using skills to help match people with employers. Probably the employers and the potential employees will not learn a new skills vernacular. They’ll talk in the way they’ve always talked about themselves and put whatever they think is important on their resumes. But we code that, figure out a way to translate that, and let the machine do its work.
A previous version of this post incorrectly stated that the DOORS application in Hawaii launched last November. It launched on March 31 (corrected 4/11/22).