Amid controversy and emerging regulations around AI for talent management, Beamery emphasizes explainability

Among the many use cases for artificial intelligence (AI) is for talent management.

Using AI for talent management and human resources (HR) purposes is, however, not without its challenges, as regulators are increasingly trying to put controls on the technology. For example, New York City is currently working on the Automated Employment Decision Tool (AEDT) law to help bring visibility and governance to the use of AI. 

Among the vendors in the space is London-based Beamery, which is continuing to build out its AI-powered talent management platform in an approach that the company’s leadership is hopeful will satisfy existing and future regulations. Beamery includes General Motors, Uber, BBC (British Broadcasting Corporation) and Johnson & Johnson among its users. 

Beamery was founded in 2014 and had an initial focus on the talent-acquisition side, helping organizations find the right staff. The company’s capabilities and AI technologies have improved over the years, and Beamery’s platform now includes a host of other talent lifecycle capabilities including skill development and mobility.

“We’ve really expanded on the product suite, really doubling down on working on the data layer around understanding people, their skills and capabilities, Abakar Saidov, CEO of Beamery, told VentureBeat.

To help support the company’s technology and go-to-market effort, Beamery announced today that it has raised $50 million in a series D round of funding. The new round was led by Teachers’ Ventures Growth (TVG).

The evolution of AI for talent management

The first generation of AI technologies for talent management were largely about matching job descriptions to resumes.

Sultan Saidov, cofounder and president at Beamery (and brother to CEO Abakar Saidov; hereafter referred to as “S. Saidov”) explained that basic pattern-matching for talent is a less-than-optimal approach to find the ideal candidate. What Beamery has developed is a much more nuanced approach that makes use of graph data models and AI to create contextual understanding. For example, he noted that it’s important to understand what a company does and what job titles mean inside of a specific company.

By having a contextual understanding, S. Saidov said that it’s also possible to better identify potential candidates that might otherwise not be found.  

“We identify the types of people that are going to be easily trained or are trainable, even if that skill set doesn’t exist today,” he said.

By having the contextual graph of how skills and requirements relate to each other, it’s also possible to help recommend career paths. S. Saidov said that Beamery can now recommend to a new hire which courses that would help them navigate to their desired career goals. 

The impact of HR regulations and need for explainability on AI

At the core of the various HR regulations, in New York and elsewhere, is a need to help make sure the AI-driven systems are working in a fair and equitable way.

A primary way in which vendors like Beamery are looking to comply with regulations is by providing explainable AI approaches. Beamery has published an explainability statement to help users and regulators understand how the Beamery platform works from a visibility perspective.

S. Saidov explained that the regulations are usually about requiring organizations to prove the AI models are auditable. The audits need to be able to identify if there is any overt bias in the recruiting or decision-making process.

In S. Saidov’s view, many of the HR laws around AI right now, including the one in New York, are still in a somewhat ambiguous state, partly because the laws are often too vague even in the definition of what AI actually does.

In Beamery’s case, S. Saidov emphasized that his company’s platform doesn’t do automated decision-making.

“We never say ‘hire this person,’” S. Saidov said. “Everything that we do is about providing explainability. For example, here are career paths you could have or, even if you’re a recruiter, showing parameters that you could consider to help you evaluate people.”