It may sound like the beginning of a bad joke, but bear with me: What happens when your company’s AI recruitment software, trained to review job applications, turns out to be sexist and discriminates against female applicants?
If you’re tech giant Amazon.com Inc., you learn from the computer science aphorism, garbage in, garbage out (GIGO), and you scrap the machine-learning tool’s algorithm. Amazon reportedly never relied solely on the male-dominant recommendations of its recruiting engine, according to wide reports that broke in October 2018. But it’s this type of story that can set back the public’s perception of AI and its nefarious ramifications (in this case, automating human biases instead of removing them from hiring decisions).
So, this whole debacle begs the question: Why is bias-free AI so much harder than it seems? Well, as society goes, so goes the business world, with an assortment of biases — unconscious or flagrant — from gender to race, religion, age, sexual orientation and elitism. All biases must be unmasked and removed when a person (or machine) is assessing an individual’s capability of performing that job, said Kamal Ahluwalia, president at Eightfold.ai.
“If your system is not designed to look at talent holistically, then you’ll run into issues,” Ahluwalia said. “If the machine learning is only predicated on past company actions, then of course you’ll end up with a different outcome than intended because that’s what you’re trying to fix.”
Mind the Talent Gap
The promise, threat and arrival of AI have been covered exhaustively in the news. But its deployment is only starting to hit full stride in the strategic acquisition, retention and engagement of talent.
While most CEOs and CHROs seem to understand that AI can help with their talent challenges, only 22% reported using AI for their talent management systems, according to the 2018 “Talent Intelligence & Management Report” by Harris Interactive Media and Eightfold.ai. The report is based on a survey of 1,007 CEOs and CHROs of companies with 1,000 or more employees.
CEOs and CHROs in the Eightfold.ai report stated that they will lose nearly half (47%) of their top talent within two years of hire. 28% of roles will never be filled, they said, and 87% of respondents said this failure hurts their company’s growth and competitive position.
With skills shortages, greater market competition and unfilled job openings, companies sorely need to develop their existing workforce to fill key roles. The time and expense of acquiring new talent alone should motivate organizations to hold on to what they have.
“The reality is that talent management is more than talent acquisition, more than internal mobility and more than retention,” Ahluwalia said. “It’s all three aspects intertwined.”
Organizations can use AI to optimize their TR programs and provide internal mobility and career planning for employees. 70% of execs surveyed believe this use of AI could improve employee retention and upskilling, according to the Eightfold.ai report.
In established companies, everything that’s new and exciting needs to be a positive for your existing employees, even those who are likely to leave your company, Ahluwalia said.
It may seem counterintuitive to shuffle or promote people who are hunting elsewhere for a job. But among the advantages of filling positions with existing talent are much shorter job vacancies (from 30 to 45 days as opposed to 90 to 120 days) and lower overall costs.
A vacant job becomes a career opportunity for everyone in the company “because it can be done at scale with AI,” Ahluwalia said.
When you give your employees an opportunity to grow and determine the next job in their career, all the right things start to happen, he said.
“Once they have a career with your company, the outcome is very different,” Ahluwalia said. “Retention is higher, loyalty is higher, they are bringing in more referrals. They now have a career in your company versus just a job.”
Diversity Through Internal Mobility
Diversity is no longer a buzzword or platitude that sounds good in front of the board of directors. Many companies have made a commitment to diversity and inclusion and have foundational policies that they’re working to bring to life within the walls of the business, said Beck Bailey, deputy director of workplace equality at the Human Rights Campaign.
“Companies understand that being inclusive and casting a wide [recruitment] net result in the best yields, the best workforce and the widest available pool of talent for their workforce,” Bailey said. “While it is for many companies a matter of doing the right thing, it is also fundamentally the best business decision.”
Succession planning too should not be limited to the top 5% of your company, Ahluwalia said. The alignment of the company’s needs and the employee’s career ambitions is an energizing force and demonstrates how much management cares about the well-being and future of the individual.
Before employees begin shopping their résumés, organizations should consider empowering them with self-service tools. One such solution is an AI-driven job portal that unlocks opportunities for employees by identifying an internal career path. Not only would the AI proactively recommend positions based on competencies and interests, but it would “mentor” the employees by informing them of the skills they’re missing (and the course[s] needed to fill their skill gaps).
Tata Communications, a large fiber optic cable network with corporate offices in Mumbai and Singapore, implemented such a job portal and met its ambitious diversity and inclusion objectives, in part, by locating qualified diversity talent from within the organization. Tata employs more than 8,500 people representing 40-plus nationalities across 38 countries, 30% of which are outside India.
By being more in tune with the hiring process and committed to diversity, Tata proved itself a “forward-thinking company that knows where it wants to go with AI that supports it,” Ahluwalia said.
The Age of Reskilling
We are no longer asking employees to “coexist” with AI so much as to “reskill,” said Marion McGovern, author of Thriving in the Gig Economy.
62% of executives think they will need to replace more than 25% of their workforce in the next five years because of advancements in automation and digitization, according to a recent McKinsey Global Institute report.
“The work will be there, it will just be configured differently, and workers will need to adapt to the new role,” McGovern said.
Organizations can use AI to optimize their total rewards programs and provide internal mobility and career planning for employees.
The implications of technological developments in the workplace are hardly a new phenomenon. What is new is the pace at which these changes are occurring.
“It used to be a new generation could be hired to absorb changes, but now companies will need to deploy their existing workforce in new ways,” McGovern said. “The reskilling need is very real.”
AI’s Expanding Role
While AI products and services aren’t yet bombarding compensation and benefits professionals, AI applications undoubtedly are changing the job candidate’s experience and essentially taking over the role of the recruiting coordinator. With time, some experts forecast penetration into the rewards space.
“It won’t be long before some of the AI apps that help onboard new hires will be programmed to handle benefits issues and questions,” said Greg Roche, senior director of compensation and benefits at a large health-care company. “That will free up benefits teams to focus on education and more comprehensive implementation of their strategic programs.”
Roche pointed out IBM’s development of the Watson Assistant for Health Benefits as an example. It is a Watson-powered chatbot that answers all the benefits questions employees might have. It ingests your benefits plan documents and then learns how to answer questions about the procedures covered by your plan, Roche said.
Described on the IBM website as “a costeffective engagement solution that understands health plan benefits logic,” the AI-powered conversational platform can be trained to act as a transparency tool for medical costs. For example, Roche said, when an employee asks about a procedure, a doctor and Watson both use claims data and other sources to give them an estimate of costs and suggestions for other lower cost options.
Based on a demo that he saw, Roche said, “It does everything I envision a benefits chatbot doing and could provide a higher level of service and benefits education to employees.”
In compensation, AI eventually will do the job of evaluating job descriptions and market matching. It even may take care of delivering the market analysis to end users without much interaction from the comp team.
“That means the comp team can spend more time coaching managers on how to talk about comp and have in-depth conversations with employees about total rewards,” said Roche, a career transition coach and self-described “non-HR” HR guy.
It may be a few years until you witness the organizationwide invasion of AI, but there’s no sense denying reality: AI’s presence will be felt.
It may be a few years until you witness the HRIS-compatible organizationwide invasion of AI (complete with nonjudgmental HR bots), but there’s no sense denying the stark reality: AI’s presence will be felt at a desktop near you.
“AI will not eliminate all comp and benefit roles, but it will change what those roles do and focus them more on human interaction with the business as a consultant and adviser,” Roche said.
This may mean that when employees question whether your recent talent acquisition wasn’t part of some skewed AI-manipulated human assessment, you can report with confidence that your company has concrete, non-GIGO data to justify its nonbinary hiring decision.