How LinkedIn is using AI to improve its job-search features

It’s an unfortunate reality of today’s job market: many would-be hires spend days scouring job descriptions and blasting applications to each opportunity they find.

That’s one reason LinkedIn, one of the most-used profile job-search sites, is implementing AI features so job seekers can discover job matches, opportunities, and relevant connections much faster. The new tools make a job search more of a conversation, with less time spent scrolling through job descriptions, Rohan Rajiv, product lead for job search and jobs marketplace at LinkedIn, said in an interview with Computerworld.

“We’re reaching a world where you can search for what is uniquely important to you — not the same keywords everyone else is using. That means less wasted time and more relevant matches,” Rajiv said.

The AI tools can establish a job seeker’s or recruiter’s intent. Those looking for work can type conversational questions into a search box and get more relevant listings. Based on a candidate’s profile, they might also receive listings for alternative careers.

The hiring experience is also easier, as AI tools surface the most qualified candidates and those with interesting backgrounds. (An OECD report in June tied labor shortages in specific industries to the inability to find skilled candidates.)

That creates real economic opportunity and efficiency with quality, not quantity, in the application and hiring process. “We’re beginning to see the Lego pieces that fit together,” Rajiv said. “First, you should be able to just say what you want and be able to get it. Second is when we have opportunities, we should be able to quickly understand if you’re a match.”

“Then third, we should be able to connect you with people who can help get your foot in the door. The recruiter, the hirer should be able to say what they want and find people.”

The cornerstone of the platform is AI job search — a major improvement over the traditional search tool, which didn’t understand intent. For example, traditional search for “entry-level marketer” returned “a mixed bag of stuff because ‘entry level’ is actually not something that we always used to understand,” Rajiv said.

The new AI-powered search can dig much deeper — much like AI-powered internet search engines — and return more unexpected results. “It’s completely semantic. So, now I can kind of have an aquatic technician, fish protection, all of this,” Rajiv said.

AI-powered job search became available earlier this year in the US, UK, Canada, Australia, India, and Singapore. A broader rollout is planned for 2026, a LinkedIn spokeswoman said.

LinkedIn this month also launched “People Search,” an AI feature that connects applicants to potential references inside a company. Job seekers can type in, “Who can refer me to Accenture” or “Who knows about diffusion models,” and discover relevant connections within the network. Those results previously did not appear through traditional searches.

“You’re suddenly getting to this world where, as a job seeker, you’re able to quickly ask a question here and find answers that you would otherwise not have found,” Rajiv said.

The human-in-the-loop feature arrives as corporations increasingly turn to AI to screen job applicants. That trend has been met with skepticism amid concerns about hidden bias. (A Gartner survey in July found that only 32% of job applicants trust AI will fairly evaluate them.)

For LinkedIn, People Search adds humans to the automated job search equation. “I want a lot of that drudgery to go away so we can spend more time connecting with humans, because that is going to be the path to opportunity,” Rajiv said.

People Search is currently available only to premium subscribers in the US. “But it will go global across all languages in the next months. It just takes time for us to … iterate and scale. It will be available for free…,” Rajiv said.

The slew of AI tools should also remove bloat and redirect candidates from poorly-fitting roles. The system shows candidates what skills they’re missing for specific positions, pushing them toward better-fitting opportunities.

“Within seconds, we’ll know, ‘Hey, what am I fit for and what am I missing’? And you can see it’s pretty clear in terms of like, it needs laboratory or field biology and research. And I don’t have that,” Rajiv said.

He said about 2 million applications are getting redirected monthly, which is discouraging low-quality matches and encouraging high-quality applications.

For recruiters, “the ability to match on both sides and to showcase that match is a lot stronger. We’re just beginning to see the benefits of rolling that out,” Rajiv said.

LinkedIn’s new AI tools are based on an LLM recommender system, which is now able to do more. “The big change is that previously these recommender systems were built on CPUs and now we are doing these systems on GPUs,” Rajiv said.

The new hiring tools mostly focus on full-time jobs, but roles such as contract and gig work are also under consideration. About 5% to 15% of the US population earns income through gig work, according to a Goldman Sachs study released earlier this month. 

Rajiv declined to go into details about those efforts until they gel.

In the future, the latest AI upgrades could allow LinkedIn to add new features as the system matures, including recommending mentors, Rajiv said.

“Imagine down the line, this adds your posts, adds your content on LinkedIn. And at that point, basically, these models can then understand, based on all your content, that, hey, you are somebody who can mentor,” he said.