The next phase of the corporate AI boom: measuring which employees actually understand it

• Companies have scrambled to adopt AI in the workplace.
• Now they want to measure how well their employees understand and use it.
• Workera's CEO says measuring effective use of AI is the next phase of the AI boom.

Executives are hyping their AI adoption rates like it's their quarterly revenue.

Over the past 90 days, CEOs have referenced "early AI adopters" at least 60 times in earnings calls, conference presentations, speeches, and other prepared remarks, according to business intelligence platform AlphaSense, signaling their determination to show they're outpacing competitors in the race to leverage AI.

Now, some of the largest American companies are shifting from simply adopting AI to assessing how well their employees actually understand it.

"If you can measure people fairly accurately, you can actually do a lot of things better in society," Kian Katanforoosh, CEO and founder of Workera, a business skills intelligence platform, told Business Insider. "You can hire better and more meritocratically. You can match people to the right project. Internally at companies, you can determine what people should learn to maximize."

Workera says it now serves about 10% of the Fortune 500, helping them understand AI and upskill their employees on everything from basic terminology to recognizing biases in AI systems.

Workera's AI fluency framework goes well beyond just knowing how to prompt ChatGPT. It's divided into three parts to assess employee fluency: AI, generative AI, and responsible AI.

At the most basic level, it asks that employees be able to differentiate between machine learning, deep learning, and generative AI. Another requirement is the ability to describe an AI agent and its basic functionalities — a concept that can feel nebulous amid the current AI race.

To evaluate generative AI fluency, employees should be able to write basic AI prompts and spot hallucinations in AI-generated responses. They should also understand how large language models are trained and generate information.

And while responsible AI sometimes seems to take a back seat at major corporations, Workera's framework asks employees to distinguish between biases in AI systems that are algorithmic, data, and human, and to explore common privacy risks in AI.

Through its initial assessments, the company has found that only 11% of users accurately estimate their proficiency before taking an adaptive AI assessment, while 32% overestimate their skills, and 56% underestimate them.

Katanforoosh said this shift to assessing AI skills is part of a broader transition in the tech world.

"The last decade in education was about access. The next decade is about measurement," he said.