Inventing a Job-Skills Machine

A free tool makes it easy to design credentials around job skills. Are colleges ready?

The problem with infrastructure is that it tends to be expensive. And slow to build. And hard to maintain. Oh, and also kind of boring. Most people only notice it when it collapses, like an old bridge.

People with typical imaginations have trouble envisioning new infrastructure before it exists. They can grow so accustomed to their daily problems that they don’t pause to wonder whether a solution is possible.

And yet every once in a while, someone has an idea, puts it to the test and invents something new—which then becomes indispensable. Like the steam engine. Or the telephone. Or the internet. Inventing a Job-Skills Machine.

Bear all this in mind as you consider the following question: What if there were a job-skills machine?

There’s a man down in Texas who thinks he’s found a way to build one. His name is Michael Bettersworth, and he’s a vice chancellor and chief innovation officer at Texas State Technical College. After years of thinking and tinkering, coding and categorizing, he and his team are ready to share their job-skills machine with the world.

The free tool, called SkillsEngine, contains a large digital library of skills—more than 20,000 of them—needed for all the jobs you can think of—and, importantly, for occupations that don’t even exist yet. The system tracks a broad range of interpersonal and technical abilities, everything from possessing empathy to clocking in with an electronic timecard. Related skills can be bundled into skill sets, and those sets can be arranged and rearranged to create skills profiles for jobs as varied as truck service technician and software developer.

SkillsEngine performs clever tasks. Feed it a typical job description or help-wanted ad—that is to say, one that’s likely either vague or full of jargon—and its artificial intelligence-informed system spits out a new version of that text translated into the language of skills. The federal government has used this to standardize its job descriptions and to help employees identify new roles that might be a match for their talents.

The system also makes it possible for a college to build a credential around the skills needed for particular occupations. A community college in Denver used SkillsEngine to create a cybersecurity degree program. Leaders in Minnesota used it to build training pathways for personal care assistants.

And that’s not all.

Spend an hour—or four—conversing with Bettersworth, and he’ll explain why he believes SkillsEngine will help companies improve hiring, workers find better jobs and higher ed institutions more successfully train students for good careers.

That last point—let’s dig into that. Because although those first two ideas are pretty uncontroversial, not all colleges or professors are eager to embrace the idea that they ought to be in the business of preparing people to fit neatly into the labor market. As opposed to, say, teaching students how to live “the good life,” or how to discern meaning through art and literature, or how to critique the very system that demands human productivity in exchange for pay in the first place.

Philosophical qualms aside, supporters of skills-based hiring and education say it’s tantalizing to think of the potential that could be unlocked by breaking jobs down into their itty bitty parts—like mapping the human genome, one coil of DNA at a time—then applying that information to engineer new opportunities to connect people with employment.

“We want to build this thing that you’d be crazy not to use,” Bettersworth says. “That’s the bull’s-eye.”

Sorting Skills

A common analogy compares skills to a currency, one that has value precisely because it can be counted in nickels and dimes instead of only in Benjamins. Right now, many employers will only consider job candidates who apply already holding an entire $100 bill—a whole college degree. But what if that large bill didn’t matter so much as the hundred individual dollars it represents, or even the 10,000 pennies? Wouldn’t a hiring manager be thrilled to find someone who is just a few dollars, or a few cents, short of fully qualified?

This is why job skills are having a moment. As the federal government, think tanks and nonprofits try to figure out what makes a good job and how to get more people employed in one, a growing chorus of voices is calling for companies to pay less attention to whether a job seeker has a full college degree and more attention to what tasks he or she can actually do and what knowledge he or she has to offer. The hope is that this would lower a big barrier blocking many people from opportunities where they could thrive if given the chance.

Sounds reasonable enough. But making that shift would require a lot of systems to change, in some cases dramatically. Employers would need to recognize which skills—inventory management? soldering and brazing? conflict resolution?—are truly needed for their open jobs, then figure out how to find people who have those skills. Job seekers would need to know how to articulate and demonstrate their skills during the hiring process. The traditional resume? Dead. And education institutions likely would need to be able to explain how their programs prepare students for a skills-based employment system.

What exactly is a skill?

Before you even get to all that, though, it would be helpful to know, well, what exactly is a skill? Is a “soft” one as important as a “hard” one? Even if everyone could agree on the answers to those questions, would we be able to create a shared language for describing all these skills, so that workers, bosses and educators really understood each other?

Trying to solve this puzzle predates the current skills-first movement. Industrial-organizational psychologists and other experts have been doing job and task analysis for decades on behalf of the military and the federal government. One long-established repository of skills knowledge is called the Occupational Information Network, or O*NET system. Housed in the U.S. Department of Labor, it has data about more than 900 occupations. Colleges use it for curriculum development and to help students explore careers, employers use it to craft job descriptions, and workforce experts use it to help people who need new or better jobs find roles that match their skills. Even other countries use it.

Creating and maintaining something like the O*NET system takes a lot of work. Compiling, sorting and validating information about which skills are important for which jobs traditionally has required gathering managers and workers together for extensive interviews, or else sending carefully crafted surveys out to workplaces. These methods yield solid data, but it’s a slow process, often lagging behind what’s happening in the job market right this minute. Inventing a Job-Skills Machine.

New system is needed

There are other ways to try to measure which job skills are in demand at any given moment, like scraping data from live online job postings, but that methodology doesn’t impress Bettersworth—he says it can miss a lot of information.

He thinks a new system is needed. A skills engine. One that hits a sweet spot, blending the rigor of traditional research methods with the convenience of modern technology. It should “avoid tormenting people,” he says, by being easy to use. And it has to clearly offer practical value to someone who sits down to use it.

“Most people aren’t interested in data standards or conforming to a taxonomy,” Bettersworth says ruefully. “We really need to avoid toiling, and overly complex solutions. Elon Musk said recently, ‘the best process is no process.’ There’s not a lot of tolerance for it.”

The team that created SkillsEngine did a lot of work on the backend so that the user interface feels, if not exactly entertaining, at least not intimidating or laborious. To develop a skills profile for a particular occupation—like personal care assistant—the tool presents a user with a simple quiz that asks him or her to rate a series of work activities—like “manage clinical case records”—based on how essential the users believes they are to the job: critical, important, beneficial, or irrelevant.

Prioritize

After enough qualified reviewers complete this quiz, the system compiles all the identified skills, which might reach into the hundreds. It’s overwhelming to think about trying to hire someone who possesses that many abilities, or to design a curriculum that teaches all of them. But the idea is that knowing which skills really matter—which are “critical” or “important” instead of merely “beneficial”—can help hiring managers and educators prioritize what to look for and what to teach.

SkillsEngine started its life as a way to help Texas State Technical College assess how aligned its programs were with employer needs. Then it became a business-to-business product licensed to colleges, states and credentialing organizations. Now, thanks to investment from the college, the project’s leaders are shifting strategies to “set the data free,” Bettersworth says.

He hopes this will attract a community of people willing to contribute their own insights to help it grow. If, for example, an accountant puts in a bit of time to review the skills profile for her profession, she will be contributing to the knowledge of everyone in the workforce-training world. It’s like how many people are willing to share their personal salary information with the website Payscale in order to get access to and improve its crowd-sourced compendium of wage data.

“Our goal is to not necessarily build the next app. It’s to build the ecosystem for intelligent, always-updated skills data,” Bettersworth says. “We are starting at the frontlines with practitioners—with tools that solve problems first—rather than starting with infrastructure and building tools around it.”

Potential to Collaborate

Pam Frugoli, a workforce analyst and O*NET and Competency Model Team Lead at the Department of Labor, thinks new systems like SkillsEngine have the potential to complement the government’s own.

“We hear constantly that people would like us to add hundreds of occupations that are more detailed to O*NET, and we can’t,” she says. “We think it’s very valuable that these other systems dive down into certain sectors and occupational specializations.”

She’s especially interested in seeing what’s possible when tools like SkillsEngine apply natural language processing to skills data, which may help to identify new, useful relationships and insights hiding in all that information.

“I don’t think we’re ever going to be able to standardize the taxonomy of skills,” Frugoli says, but then adds, “I think we could get a better handle on it with artificial intelligence.”

Right-Sizing Education

When Texas State Technical College set out to reevaluate its associate degree program in web development, a curiosity emerged. Why, reviewers wondered, did students have to take three semesters of classes about copyright law?

The requirement turned out to be a relic from the days when colleges turned into intellectual-property crime scenes thanks to the popularity among students of the music-streaming service Napster.

But by the 2010s, teaching so much copyright law to future web developers seemed unnecessary. After all, that information was unlikely to help them find jobs. Compared to all the other material students could have been learning, copyright law was, as Bettersworth puts it, “over-indexed.”

So the program shrank the copyright courses down to a more reasonable size: a mere module.

“The greatest injustice to a program like that—one lacking market relevance—is it’s an impediment to the students’ employability,” Bettersworth says. “You’re wasting a person’s time, precious time, money, their hope, in taking a curriculum that may have been relevant 10 years ago but isn’t today.”