Artificial intelligence begins driving the employment process

Artificial intelligenceIn theory, removing bias from the employment equation should enable employers to hire and manage the best employees based on knowledge, skills and abilities. Several HR-related artificial intelligence software packages attempt to do just that.

If the hype is to be believed, AI should herald the dawn of a perfect meritocracy at work.

THE LAW Numerous federal and state laws govern hiring, promotions, recruiting and just about every other phase of the employment process. The Civil Rights Act of 1964, the Age Discrimination in Employment Act (ADEA), the ADA and the Genetic Information Nondiscrimination Act (GINA), all carve out anti-discrimination protections for specific groups.

WHAT’S NEW Artificial intelligence promises a streamlined workplace where software reads résumés, compares them to job descriptions and sifts a short list of well-qualified applicants from a sea of online résumés. It promises greater objectivity and therefore expertise than managers could ever bring to the process.

For example, a hiring manager may recognize a Stanford IT degree, but may not know which Chinese or Indian schools produce the best programmers. Some AI packages will identify prime candidates even if they haven’t applied to the company.

Hiring for Attitude D

Promotions can be handled in the same dispassionate manner, letting the true stars shine despite office politics. Similarly, managers no longer have to feel bad sitting across from an axed employee: “So sorry, the software made the decision.”

Several AI products promise to increase retention by comparing employee emails to algorithms that distinguish between disgruntled employees and those might simply benefit from more challenging assignments. Software can help determine if either employee is worth working hard to retain.

Software can also be used as a policy enforcement tool. It can track computer usage and report how much time an employee spends browsing social media or shopping.

HOW TO COMPLY So, what could possibly go wrong? First, it’s relatively new tech. Most algorithms are based on past experience, which may pull from a small sample size. Second, AI collects lots of data. Not every manager will know how to use it.

AI in the hiring process

While AI should help eliminate hiring bias, it’s a mistake to rely solely on algorithmic interpretations. For example, an outdated job description will send the AI program searching for unqualified applicants.

Employers should be able to tweak search parameters so the software identifies the candidates with the right knowledge, skills and abilities for the job. For example, to target candidates who currently or have in the past worked for a specific competitor, the software should be able to adapt to changing market conditions.

Also note that raiding a competitor with a history of discrimination may replicate that employer’s legal headaches along with its workforce.

Workforce management

Although, employers are generally free to monitor employee use of company computers, some caveats apply. For example, state laws may limit the data employers can capture, such as social media passwords.

Similarly, some activities that employees perform on company time may not be work, but may be protected under the National Labor Relations Act.

Finally, focusing on what employees do online may not actually measure what they are producing.

An extremely efficient employee could spend substantial time surfing the web because she quickly finishes all her work. Looking only at her nonproductive time may result in disciplining an employee who is simply not challenged by her job. A promotion may be a better move.

Software that allows employers to track employee movements can be extremely useful when tasks must be performed in a particular sequence. It also can help employers limit access to specific physical spaces when necessary, such as secure areas. Systems like these require employees to wear badges that transmit their locations. Employers may wish to auto­­­matically deactivate badges when employees leave work so as not to violate their privacy.

Enhancing employee retention

Because there is no one holistic AI program, employers must be educated consumers of AI. Managers must know what the AI data means.

While some packages are meant to monitor productivity, they may not detect signs an employee is bored at work. Software that focuses solely on retention may send managers on the fool’s errand trying to retain a worker who just isn’t producing.

AI has made many market-driven, task-specific advances. However, judgment, at least for now, remains a solely human skill. Employers must understand each software package’s scope and limitations. Managers using AI data must be trained to un­­derstand all its implications.

Employee notification

Because AI software essentially monitors employee behavior, employers should have employees sign waivers stating that they understand they are being monitored in the workplace.

Note: When first introduced, such surveillance may produce a “Hawthorne effect,” in which employee performance is higher than normal. Employers should wait until the novelty has worn off to establish baseline readings.