While AI applications and recruiting seem like a natural fit, real world experience is proving the combination might not be quite ready for prime time.
Reuters news service reported last week that Amazon recently shut down an internal R&D program for an AI Recruiting tool, after it was discovered the tool had evolved to discriminate against women through a variety of assessment factors.
According to the report, executives lost confidence in the program even after the discriminating assessment factors were purged from the model, as they couldn’t be sure other hostile factors wouldn’t be reintroduced in the future.
With 55 per cent of hiring managers saying they intend to use Artificial Intelligence in their work within the next five years, what should organizations know about it right now?
AI still lacks understanding of what recruitment really is
The issue of biased selection models is just the start. The core factor concerning AI and recruitment is a misunderstanding of what recruiting actually is. At a surface level, the recruiting process seems like a matching and categorization exercise, both of which are broad types of things that machine learning techniques does well in other areas, like online retail product recommendations and search engine results.
But when you peel back the layers, there is more going on. Good recruiters work with their clients to help them understand what they are truly looking for, and then go hard at work in qualifying likely candidates – an exercise in emotional intelligence that AI tools lack the capability to perform.
AI automation can be successful in the basics of recruitment, but there are gaps in other phases of the staffing lifecycle
Job descriptions and candidate resumes are important starting points in a recruiter's work but are also flawed data points for different reasons. Until AI can grapple with and resolve those imperfections, industry executives believe it's likely that AI recruiting uses will be narrow and focused on specific tasks in the overall recruiting process.
Forward thinking staffing organizations that have applied machine learning to their core sourcing activities have experienced success; however, other phases of the cycle, like client engagement, candidate qualification and closing are still people led, supported by a mix of non-AI technologies.
It’s important to be aware of AI—but not hasty!
Here at Procom, 2018 marks our 40th year in business, and that time has marked a profound explosion in how technology has shaped recruiting. In the early years, paper resumes ruled, and at one point, even faxing felt revolutionary. Yet today, after auditioning over 50 commercial products in the last 12 months, we have formed a decent appreciation of the challenges the current generation of AI solutions face.
Today, recruiters have countless paths to source and qualify great candidates, automate necessary but low value work and drive better service to clients. AI methods will undoubtedly be an expanding part of the professional recruiter tool set, but the vision of full AI recruiters parsing applicants and making the ultimate hiring determination is a distant promise. At least for now.