Microsoft Patents AI-Generated Talent Screening for Hiring
Microsoft filed a patent for a new AI in HR algorithm in early March, a move that should come as no surprise for a brand with more than 77,000 patents globally.
The new patent, flagged by Patent Drop, includes a collection of methods, systems and computer programs that automatically generate phrase-based talent screening questions, promising to simplify the hiring process.
Microsoft: Innovation Necessary in Job Market
“When placing a job posting, job posters sometimes add screening questions for potential candidates in order to simplify and accelerate the selection process for candidates applying for the job,” Microsoft stated in its filing.
Figure 1 from Microsoft patent filing.
Some solutions for this process include machine learning, the company added, but this path requires manual labeling of a training set, which can be very large and rely on a significant amount of human labor.
Other solutions may use word searches to identify a job requirement, but they’re subject to false positives. For instance, a drug test development company might create a false positive for a requirement that employees submit to drug testing.
“As the job market evolves, so do the requirements for new jobs,” Microsoft stated. “The changing needs of the market means that new screening questions will arise.”
The problem with existing recommendations, the company added, is that talent screening questions are predefined, and the system selects from those predefined questions. “However, if new trends in the job market require new screening questions, existing recommendation systems are not able to automatically modify the questions.”
Related Article: Are You Safeguarding Against Bias in Your Hiring Systems?
Microsoft Patent for AI Generation of Talent-Screening Questions
Microsoft’s newly patented solution, an AI-backed algorithm, works in two parts:
First: It analyzes job descriptions to generate ngrams, or sequences of words, that indicate job training requirements. For example, it might pull out ngrams like “valid driver’s license” or “authorized to work in US.”
Second: It matches those extracted key phrases to recommended new talent screening questions. Based on the ngrams above, the questions might be “Do you have a valid driver’s license?” and “Are you authorized to work in the United States?” The job poster then has the option to include those questions or not.
Figure 2 from Microsoft patent filing, which depicts part of the keyword phrase discovery and extraction process.
Microsoft declined to comment on how it plans to use this patented technology. On the surface, however, the most obvious prediction would be an integration with LinkedIn (shown below), a company Microsoft acquired in 2016.
LinkedIn, like most job board websites, has both free and paid job posting options, with paid tiers including advanced features like promoted postings and instant mobile alerts to qualified candidates.
This newly patented AI tool could be the value prop LinkedIn (and Microsoft) needs to attract the 25% of organizations that plan to start using or increase their use of automation or AI in recruitment and hiring in the next five years.
Related Article: Artificial Intelligence in HR Remains a Work in Progress
What Comes Next for Artificial Intelligence in HR
Microsoft also declined to speak on if or how bias — a common challenge that appears in AI applications — was addressed in the development of this technology. Though just days later, it officially announced the layoffs of all members of its AI ethics and society team.
Manish Raghavan, professor at MIT Sloan, said that this application of AI in HR seems relatively benign. “As long as the questions are based on requirements that humans have decided are meaningful and relevant to the job at hand.”
Learning Opportunities
AI in HR has the potential to be positive, he added, for both firms and for candidates. “But there are a lot of ways this can go wrong without the appropriate guardrails.”
He explained one of the biggest challenges when introducing AI to a process: “AI is often used to learn correlations and patterns that humans don't fully understand or can't explain. This can lead to tools that don't work particularly well, or that replicate patterns of biased decision-making.
“Evaluating AI tools in the HR context is particularly challenging because there often isn't a clear objective benchmark of ‘candidate quality’ that we can compare AI's performance to.”
Still, Raghavan added, he believes there’s a lot of opportunity for AI in HR, “as long as the goal is to support human professions, as opposed to replacing them.”
Related Article: The Complicated Relationship Between AI and Human Resource Management
Other Notable Microsoft AI News in 2023
2023 is clearly a big year for Microsoft and AI, following the January announcement of its $10 billion investment in OpenAI. Microsoft's efforts extend beyond AI in HR applications.
In fact, as of March 2019 (the last time reported), the tech giant had more than 18,000 AI patent applications, more than any other company worldwide. Number two on the list, IBM, had just over 15,000.
Another notable Microsoft AI patent for a self-learning digital assistant dropped March 16, 2023.
“People are increasingly interacting with computing devices and relying on these devices for information, recommendations and other services to assist them in their day-to-day tasks,” Microsoft wrote in its filing.
However, the devices currently available, according to the tech giant, are unresponsive to user commands because they don’t know how to carry out tasks associated with these commands.
Microsoft’s patent outlines the specifics for a digital assistant with built-in machine learning — meaning it will learn, determine user preferences and improve as it's used.
It’s clear that Microsoft is in the AI competition for the long haul. We’ll likely see more patent drops from the company throughout 2023 as competitors release AI-backed products and the technology continues to evolve.
About the Author
Michelle Hawley is an experienced journalist who specializes in reporting on the impact of technology on society. As a senior editor at Simpler Media Group and a reporter for CMSWire and Reworked, she provides in-depth coverage of a range of important topics including employee experience, leadership, customer experience, marketing and more. With an MFA in creative writing and background in inbound marketing, she offers unique insights on the topics of leadership, customer experience, marketing and employee experience. Michelle previously contributed to publications like The Press Enterprise and The Ladders. She currently resides in Pennsylvania with her two dogs.
Connect with Michelle Hawley:
Featured Research
Buyer's Guide
Read Now
Research Report
Read Now
eBook
Read Now
Research Report
Read Now
White Paper
Read Now
Research Report
Read Now
Guide
Read Now
Guide
Read Now
Related Stories
Entries for the 2023 Reworked IMPACT Awards are now Open
