Related: LGBT employees and the workplace

The new diversity challenges are both old (racial imbalance) and new (AI prejudice). Addressing these challenges cannot be accomplished by simply checking off specific benefit options or hiring practices on a list.

With employers still engaged in hand-to-hand combat over top talent, they are looking to brokers for help. Benefits packages offer a key competitive advantage in recruiting and retention. But if brokerages themselves lack true diversity, they are immediately at a disadvantage in providing this service.

The false promise of AI

Artificial intelligence tools and platforms promise to facilitate the hiring process by sorting through job candidates and popping the best to the top of the list. And just like their employer clients, brokers often depend on such tools to identify top candidates. But if the AI-driven tools they depend on to automate their own hiring process are bug-ridden by prejudice, they are unwittingly encouraging the historical white male dominance of their businesses and industry.

Arran Stewart, founder and CVO of Job.com, is targeting this baked-in prejudice in his company's emerging blockchain recruitment platform. A veteran of the online recruitment industry, Stewart says rooting out the biases built into code by white male programmers is the only way to truly level the hiring playing field. Currently, he says, research shows that such bias robs employers of 5 percent of top performers, whose names are weeded out by AI systems due to programming language. The built-in biases can filter out women, non-white race candidates, and those with cultural differences.

"This is a big deal. It's a troubling problem. I have three daughters. How will they find their next job, if they have to go through one of these systems? It's up to me to make sure we create the fairest environment possible."

Stewart says eliminating all bias, either from human or AI recruiting, isn't possible. But machine learning systems can be improved simply by employing a diverse programming team that can diversify the language of the machine.

"AI is built for males in recruiting. It's an inherited bias. It's no one's fault. Programmatic language looks for certain words due to its inherited bias. Brokers are historically a very white male dominated market. They are also the ones writing the job specs. So the bias toward white male responses will be even greater in that industry."

Felipe Barganier, an African-American broker/consultant, has battled such language-based discrimination, written and coded, his entire life. The CEO of GAB International LLC says it begins with standardized testing in schools and spreads through society from there.

"The artificial intelligence bias is almost like an inadvertent form of discrimination. When employers look at things like credit, it's the same thing. In a lot of minority communities, there aren't a lot of homeowners. But rent doesn't show up on credit report. They may have excellent rental records, but that is not reflected in the report," he says. "As the U.S. gets more diverse, that has to be spoken to. Yet I know several large insurance companies that have no minorities in leadership positions."

To address this, Barganier is working with a team to develop a tool that, like Stewart's, would begin to balance the ingrained bias in artificial intelligence systems.

Harnessing AI for better benefits

As Barganier and Stewart attack the built-in bias of AI, entrepreneurs like Rachel Lyubovitzky are harnessing its power to address a diverse workforce. Her team is designing AI-driven systems to enhance the enrollment experience for all employees.

Lyubovitzky is co-founder and CEO of EverythingBenefits, which offers a range of services (both directly to employers and through resellers and brokers) to support human resources. Its enrollment platform guides employees through a series of queries designed to direct them toward the coverage options that best match the individual at their current stage in life. The system is designed for today's highly diverse and multigenerational workforce, where the blend of ages as well as cultural, racial and gender differences requires a benefits package suitable to the layered needs and preferences of all plan members.

Often, Lyubovitzky says, employees believe they are choosing the best coverage options when in fact they are locking themselves into a plan that's too expensive or doesn't offer what they need.

"Humans all want to make choices and be in control. It's one thing when you are picking an outfit or a movie, but with benefits, it directly affects your life. And humans are notoriously bad about these choices," she says.

"AI looks at other peoples' decisions, millions of people. AI gives you the power of multiple experiences, and the inferences they have made from those experiences. And it says, 'Here is the best one for you.' AI has the power to help you make the right choice the first time, to gauge an individual's true situation," she says.

The platform stores and shares past claims information with plan members to improve choices every year. "We can use information from last year's experience to help you find a better plan this year. The data says, 'This is what you thought about yourself a year ago. This is what you thought would work best. Here's the claims evidence.' Now perhaps you can make a better choice."

The human factor

The broker business, especially at the high-earning end, has been historically dominated by white males. Research by government and non-government organizations alike have found African-Americans and Latinos to be underrepresented compared to their percentage within the overall U.S. adult population. In addition, women continue to comprise less than half of all insurance employees, while all minorities are underrepresented in leadership positions.

Decades of biased programming that lurk behind the online employment forms prospective employees fill out helps to explain the ongoing lack of diversity. But there has clearly been more behind the lack of diversity than computer-programmed bias.

As workforce diversity expands in all directions, the need to reflect that diversity within the benefits brokerage community becomes more crucial. But getting a foot in the door as a non-white male is still difficult.

Barganier says that his ascent within the industry to a high-profile position has been marked by two patterns: few encounters with other brokers of color; and an increasing number of requests for guidance and advice from brokers of color after GAB was recognized for excellence by industry groups.

"That put a huge spotlight on me," he says, referring to awards he received from the state of Georgia in 2016 (Insurance Industry of the Year Award) and BenefitsPRO (Broker of the Year Finalist) in 2017. "People of color were coming to me before. Now, that has increased tenfold."

Barganier tells everyone that diversity within the industry is good for business.

"I believe that lack of diversity is a barrier to brokers getting certain types of groups," he says. "When you don't look like the people you are representing, that's a drawback. When you hire a diverse staff, it allows you to understand other ethnic groups and cultures. You know better how to meet their needs."

A big responsibility

Bias, prejudice, discrimination. Call it what you will, and assign blame proportionally to humans and machines. In the end, what matters is whether employees believe their employer's benefits packages says, "I respect you."

Banishing the bias is a process, not a project. And, according to Barganier, it all starts with the broker.

"You have to focus first on the broker community, because that community controls benefits," he says. "If you change the mindset of a fair number of brokers, you can change the realities within the employer groups they represent. It takes the broker and the employer working together. But the broker has to change first."