AI has made headlines for the last several years as the greatest innovation since the steam engine. But what exactly does that mean for recruiting teams?
You are likely familiar with AI’s many accomplishments in the consumer world: Amazon Alexa, Google Maps, and IBM Watson on ‘Jeopardy!’, to name a few. But beneath the hype of consumer AI, artificial intelligence has been quietly making strides in the enterprise.
Consider these examples of AI in the enterprise:
- JPMorgan Chase performs legal checks of commercial loan agreements with AI, freeing up 360,000 hours of legal review per year.
- The Associated Press automates earnings reporting, publishing 4400 AI-written stories per quarter.
- Hilton uses AI to evaluate candidate interviews, improving interview:hire rates 40% while decreasing time to fill 90%.
In 2020 it’s clear that AI is here, and it’s here to make an impact. But like any potentially transformative technology, it’s surrounded by hype, misinformation, and mixed messages.
AI in Recruiting: It’s About Prediction
The potential for AI to make an impact in all domains is huge, and recruiting is no exception. Talent acquisition succeeds when it predicts the best candidates for a job, and builds the relationships that convert those candidates to employees.
AI has the potential to scale and manage a large portion of this prediction work. At its best, artificial intelligence offers up great talent for recruiters and hiring managers, surfacing the best candidates regardless of work history, educational background, and demographic.
This is important to keep in mind when starting a conversation with any AI vendor. It might be tempting – and interesting – to discuss the technical nuances of a vendor’s AI solution, but this shouldn’t be the focus. Instead, you should focus on the impact a solution could have on your recruiting.
Properly implemented, AI can deliver dramatic improvements in quality of hire, time to fill, new hire diversity, and other critical recruiting metrics.
The Artificial Intelligence Landscape
Applications of AI in recruiting fall into 9 major buckets:
- AI-Driven Assessments. AI-driven assessments are pre-hire assessments which leverage AI to evaluate candidates faster, more effectively, and in a candidate-friendly way. These most commonly take the form of game-based assessments and video-based assessments.
- Candidate Rediscovery. Candidate rediscovery tools analyze your existing database of candidates to “rediscover” those who might be a good fit for your open requisitions.
- Job Description Optimization. Job description optimizers are similar to Grammarly, but for job descriptions. They provide wording and phrasing recommendations that make a description more inclusive.
- Ad Automation. Ad automation AI places and tests your job ads on a variety of different platforms, optimizing your ad spend.
- Job Market Forecasting. Job market forecasting software gives insight into available pools of talent for different job types, experience levels, or location.
- Candidate Relationship Management. Candidate relationship management software can leverage AI to deliver a higher level of personalization to candidates, and re-engage candidates who applied previously.
- Chatbots. Chatbots provide a convenient interface for candidates to find and apply for jobs.
- Resume Filtering. Resume filtering tools evaluate candidates’ resumes and applications to make broad-stroke screening decisions.
- Social Candidate Discovery. Social candidate discovery software scrapes social and other online platforms to surface passive candidates that may be a good fit for open requisitions.
Within each bucket are a large variety of solutions that promise similar results. While the majority of these have at least one unique use case where they excel, a select few excel in many domains.
This puts recruiting leaders in a dilemma. How do you vet vendors when they all promise the same improvements in time to fill and quality of hire? Like we mentioned previously, it comes down to impact. Which technologies are most likely to fundamentally change your recruiting strategy, and deliver transformative – rather than iterative – successes?
The Question that Cuts Through the Hype
While some AI solutions are the best technology has to offer, others are just previous generations of automation rebranded. This isn’t to say they can’t be useful. A classic example of a previous generation of automation is the spreadsheet, and few will argue the usefulness of the CSV.
That said, the highest potential for impact lies with today’s best-of-breed technology, not rebranded tech from the 1980s.
While not 100% precise, you can use this question to test whether a vendor is leveraging best-of-breed, and delivering the highest potential for impact:
“Could a team of interns, armed with calculators and flowcharts, do what this AI claims to accomplish?”
If the answer is “no,” chances are the vendor you’re talking to is making the most of today’s available tech. The reason this works is because today’s AI offers the ability to embed expertise – not just rote tasks – into an algorithm.
5 Table Stakes Criteria to Evaluate AI Vendors
While potential for impact is the most important metric when evaluating AI vendors, it’s not the only criteria. Some might promise the moon, and ultimately be unable to deliver. Use these 5 table stakes criteria to ensure you don’t end up in a disappointing partnership.
1) Experience with the recruiting space.
Recruiting is a highly regulated discipline. Inexperienced vendors run the risk of putting their customers in jeopardy.
Vendors who have experience in the recruiting space typically have a lengthy history of success, use widely accepted and validated standards when building AI (like the EEOC’s adverse impact), and understand and have demonstrated the legal implications of their technology on hiring.
2) History of success with organizations like yours.
While the number of documented AI success stories grows by the day, many applications of AI remain theoretical.
When a vendor has a library of documented, referenceable success stories from organizations like yours, it indicates that you’ll be a partner (not a beta tester), and have a built-in playbook for success.
3) Data used to identify and evaluate candidates.
With a few exceptions, most applications of AI in recruiting are designed to uncover great candidates.
Sourcing tools, whether through social media scraping or candidate rediscovery, evaluate
passive job seekers who have yet to apply. Chatbots and resume-filtering tools evaluate applicants at or during the application phase. AI-driven assessments evaluate applicants post-application. The data that feeds these tools plays a large role in their predictive accuracy and potential for bias.
You can evaluate a vendors’ ability to predict what they say they can predict by the data they use to make the prediction. Demographic data offers the least predictive accuracy, resume data offers low to medium predictive accuracy, and job-specific data offers the highest level of predictive accuracy.
4) Commitment to ethical AI.
This is also the time to look inward and decide if AI ethics are important to you and your organization. If they are, you should take the opportunity to evaluate if vendors have the same commitment.
Vendors that are serious about ethical AI have documented ethical principles, can explain how they test their algorithms for bias (and remove it if necessary), and have an external technical advisory board.
5) How a vendor audits AI for adverse impact.
One of the biggest promises of AI is its ability to make hiring more objective and fair. That said, AI is like any powerful technology: improperly built and tested, there is the potential for harm.
Vendors should be able to provide full documentation around their process for mitigating bias.
Any algorithm in the recruiting space should be vetted to prevent adverse impact against protected classes.
The Incredible Potential of AI in Recruiting
The potential of artificial intelligence in recruiting is tremendous. But the potential is accompanied by hype, misinformation, and market noise that makes it difficult to discern the applications of AI which are truly transformative. So while some early adopters are seeing incredible advances in their recruiting processes, platforms, and metrics, most TA teams remain on the sidelines.