AI Ethics Codes: Why You Shouldn’t Make Remote Work Compensation Decisions Without One
Clearly, working from home is no longer a “perk.” Consider what happened after the CEO of Washingtonian Media penned an opinion piece in May of 2021, which stated that employers have “a strong incentive” to convert remote employees into contractors.
The day after the piece was published, Washingtonian staff went on a day-long publishing strike, and took to Twitter with the message: “We want our CEO to understand the risks of not valuing our labor.”
As this cautionary tale shows, a purely clinical approach to people can have severe repercussions, and employers need to tread carefully in the world of hybrid work. New questions are cropping up daily: If an employee moves from Boston to Memphis, should they get a pay cut? Will key people in Tulsa stage a protest when colleagues with Silicon Valley wages become their neighbors?
Indeed, hybrid work is upending traditional compensation strategies, and employers need all the decision support they can get. Wages traditionally reflect the cost of living, which varies by geography (that’s why accountants in New York City earn an annual mean wage of $102,090, but in Jefferson City, Missouri, they earn $56,800). With a lot of jobs becoming untethered from the office, however, many companies are shifting away from a purely pay-by-location model, adopting a national pay median for certain roles instead.
As the rules for employee pay change, the use of AI for pay decisions is set to rise.
Employee pay decisions: Keeping humans in the driver’s seat
AI is great at dealing with a lot of math, something every organization will contend with when setting a new pay strategy. It can also help align the decisions of individual people managers with a broader framework for compensation that is designed for equity and fairness.
But this does not mean robots should take over when it comes to pay decisions. Compensation is at the heart of whether an organization is perceived as fair, and people take matters of pay very personally. An AI, while good at narrow repetitive tasks, will miss important nuances, such as an individual’s unique context within the organization and their potential strengths.
Handing pay decisions over to a robot is a slippery slope into Orwellian territory. But AI can play a critical role in helping organizations set pay decisions responsibly—leaders and managers just need to use it wisely.
The AI ethics code: A must-have for every organization
People have been mulling over AI ethics long before the technology was a part of everyday life. (Science fiction writer Isaac Asimov first introduced an ethical code for AI systems way back in 1942.)
Setting AI ethics standards within organizations, however, is relatively new. As AI is quickly becoming a business imperative, more and more experts are calling for ethics standards. This involves, for example, setting out clear guidelines for the types of data that will—and will not—be fed into the AI model, or establishing acceptable levels of AI model performance and the frequency with which results need to be re-validated. Without these standards in place, well-meaning business leaders will fall into the habit of doing what is easiest or most efficient—and then have to deal with unintended consequences, like playing out historical biases, in the long run.
While it’s difficult to measure just how many organizations currently have ethics codes in place, forward-thinking leaders are making them a priority. The good news is that there are more resources cropping up every day. The “Framework for Promoting Workforce Well-Being in the AI-Integrated Workplace,” for example, was published by the Partnership on AI in 2021 and covers everything from human rights to intellectual wellbeing to financial well-being. Such frameworks are useful guides in setting ethical standards.
There is also a buzz about new roles with titles like “head of AI ethics.” The ideal candidate for this role would have technical, regulatory, business, and communications skills. But these people are hard to find, and as an alternative, some companies are partnering with academics at their local universities to help them deal with ethical dilemmas.
Within a broader context of ethical decision making, organizations can ensure they leverage AI in the right way to make the best pay decisions for their business—and their employees.
How human-centric workflows boost AI performance
Technology vendors are finding new ways to keep humans in the loop while helping businesses yield the benefits of AI.
Consider the standardization of job titles as an example. It’s become the norm to assign different job titles for similar types of work, which creates inconsistencies. This means that when a manager logs on to a site like Salary.com and plugs in the titles of their direct reports, they can get a false comparison.
AI is a powerful tool for standardizing job titles. It can present the manager with a market-facing descriptor so that whether they are dealing with external or internal data, they are getting an apples to apples comparison. This can be fed through a platform that then presents the information to the manager so they can make more informed decisions.
But the machines can’t do this task completely on their own. To ensure that new, standardized titles are as accurate as possible, Visier has built in a human-centric corrections workflow. The workflow allows an actual person (who has niche-specific business expertise) to review the matched standard job titles, apply their real-world knowledge, and reassign any that are not sufficiently accurate. The algorithm learns from the corrections, so that the standardized labeling becomes more and more accurate over time.
The best AI-driven pay recommendations also give humans the final say. Some people think that building an algorithm that can spit out a specific number for a salary is the ultimate goal. But this doesn’t allow for enough flexibility in considering factors such as a person’s depth of knowledge or rare but valuable skill sets. Systems that present the manager with a pay range, instead of a specific dollar amount, allow for this flexibility in decision-making.
AI should guide—not rule—pay decisions
The labor market is hot and employee expectations have changed. There is a need for fast, clear guidance for managers who are making a lot of pay decisions about their direct reports, and this is where AI can bring tremendous benefits. But amidst the enthusiasm for the powerful technology, employers also have a responsibility to use AI consciously. By building robust ethics codes and keeping humans in charge, organizations can make this a reality.
Be the first to know!
Never miss a story! Get the Outsmart newsletter.