Skip to main content
U-M Poverty Solutions Logo U-M Poverty Solutions Logo


Back to faculty research

Improving Employability via Physical Crowdsourced Tasks

Carpenter with apprentice in woodshop

The project: Digital technologies that support employment — like LinkedIn and Massive Open Online Courses — are primarily used by individuals with higher education levels. According to a 2015 Pew Research report, American job seekers with lower educational attainment would benefit the most from using digital tools to support their employment endeavors. To understand opportunities to develop these digital tools, this study examined how best to connect underserved job seekers with available work in the gig economy that could increase their employability and help them re-enter the job market. Underserved job seekers include people living in low socioeconomic areas, people who have low income and people who have limited education. They may not feel comfortable using technology to connect with employers, may not fully understand the skills needed for employment, and may not know how to clearly articulate the skills they have. The study explored effective uses of crowdsourcing to compile a list of tasks employers want completed, how to match underserved job seekers with those available jobs and what it would take to translate a series of tasks into long-term employment.

The process: In phase one, the study used Craigslist and other low-barrier gig economy platforms to compile a list of more than 400 tasks, which were labeled according to the skills needed to complete each task. The study then created mock “baseline” resumes typical for entry-level workers and mock “enhanced” resumes that included more gigs/skills to see which type of resume received the most callbacks from employers and which specific skills elicit callbacks.

In phase two, the study explored the use of online platforms to link tasks together and help workers build skills that could lead to higher-paying jobs, while earning income. The final step was to connect the online skills-building platform with the list of available gigs.

Results: The study piloted a job-search tool called DreamGigs that provided job seekers with a list of local available jobs and volunteer opportunities. DreamGigs specifies which skills job seekers need to develop in order to achieve their ideal jobs and identifies available opportunities to help them acquire those skills. The study found Craigslist lists plenty of needed tasks and does not require much digital literacy, but low-resource job seekers did not feel safe exploring jobs on Craigslist.

In terms of using an automated system to help workers build skills, the study found a series of challenges: most tasks are only available once; descriptions of tasks vary widely, making it difficult to match keywords; the way tasks are sorted by existing platforms assumes all workers are uniformly inexperienced or only shows jobs suited for their current skill set, which leaves little opportunity to build skills; and variation in pay makes it difficult to outline a path for workers to increase their pay rate as their skills progress.  

“Despite this, we see potential paths forward in using machine learning to learn more complex patterns, allowing it to help workers strategically bridge work opportunities to build and evidence their skills,” the study’s final report states.

More information: DreamGigs: Designing a tool to empower low-resource job seekers

Tawanna Dillahunt, U-M School of Information
Walter S. Lasecki, Michigan Engineering, Electrical Engineering and Computer Science