How might we help works measure soft skills when applying for gigs or employment? How might we help employers find people for tasks/contracts/jobs, who will be trustworthy/personable/good cultural fit?
This was the prompt my 4 person team was asked to think about during our week long fellowship at IDEO’s CoLab.
What we ended up with was a tool called Omiiko (which means empathy in Igbo). Omiiko uses multiple AI systems to create a reliable soft-skills profile, which can be used by workers to market their skills, and by employers to find the right cultural match for their jobs.
We built a prototype showing both the frontstage (worker) experience, and the backstage (employer) experience. The worker experience is a very simple data-entry interface answering questions while recording their video response. The employer experience combines face-capture based sentiment analysis (using Affectiva) and personality assessment (using Watson Personality Insights). We also built a demo of the interface which allows an employer to access a database of candidates and search by soft-skills profiles generated by the AI systems.
The gig economy is powered by soft skills. There is no structured way of presenting and marketing these soft skills to employers. For workers, the key benefit is the ability to capture and highlight these unique skills with minimum effort.
Recruiters and employers can easily assess hard skills. But assessing soft skills is a major overhead. This concept gives them a head-start by letting them search for candidates by soft sills and “superpowers”.
Among other things, what we learned from developing this tool is that employers love any and all context when recruiting for and filling positions. We learned that inclusivity is a challenge in analyzing soft skills or expressions.
While working with these technologies, one of the issues we had most difficulty reconciling were the biases that had been pre-baked into the models we were leveraging to analyze individuals. An individual or team is initially responsible for determining what a model defines as ‘agreeableness’ or ‘conscientiousness’. This subjective definition is almost impossible to fully agree upon person to person, so the designers of the systems may now have far more power in inadvertently deciding which soft skills are perceived in what way. Furthermore, the recruiter/hiring manager would now be responsible for reporting if they felt the systems analysis of the individual was accurate, thereby infusing their own biases into the model when it runs next.