The annoying thing about the future, a chief financial officer might think, is that preparing for it costs money—and often an awful lot of it. At this stage of AI development, businesses are spending most of their money on hiring people, not building bots. In a bleak employment landscape, every little bit helps.

“The broader macroeconomic uncertainty we’re seeing continues to play out in the labor market, which is stuck in a bit of gear,” said Sue Duke, LinkedIn’s Head of Global Public Policy and Managing Director for EMEA. “Hiring is slow. The momentum is not there. For the most part, in advanced economies, we’re seeing hiring about 20% below where it was pre-pandemic.

“One of the really bright spots is AI-enabled jobs—what we call ‘new collar’ jobs. This is a new category of workers that brings a combination of different skills, mixing advanced technical skills with clear human skills to create these new roles.”

Globally, between 2023 and 2025, about 1.3m new roles will be added in the new collar category. Data annotators, forensic analysts, and forward-deployed engineers are roles dedicated to preparing businesses for future technology and implementing AI changes. Others are easier to understand—Heads of AI and AI Engineers do what they say on the tin.

“These are roles that you didn’t hear about two years ago, a year ago, maybe six months ago, and yet we’re seeing it explode on the platform,” Duke said. “That tells you that this new digital economy, this transition to an AI-driven economy, continues to create these new categories of roles and workers that we’ve never seen before.”

Many will heave a sigh of relief upon hearing this—which may be premature (IBM it was recently announced that artificial intelligence assistants now handle 94% of routine HR tasks). Duke also told me that two-thirds of jobs will undergo fundamental change by the end of the decade.

“We expect, from looking at our own data, that 70% of the average skills of the average job will be changed by 2030. We know that we need to focus more on re-skilling and lifelong learning than ever before.”

No wonder ‘heritage marks’ are dropping applicants’ resumes. Who needs to know what school an applicant went to twenty years ago when ‘AI coding skills’ and understanding ‘token sequences’ are now more important considerations.

“Traditionally, we’ve relied on legacy signals,” Duke said. “We ask questions like: ‘What school did you go to? What degree did you get? What was your last job? What was your job before?’ What needs to change is to move away from only relying on signals to ask the most important question: do you have the skills and potential to do this job?

Understanding what skills are, and who has them, is the new power of workforce planning. The number of applicants per job has doubled since 2022, according to LinkedIn data. Employers are using AI tools to screen candidates. Candidates are using AI tools to figure out how to beat this strain, an HR arms race that’s often unhelpful—and frustrating, on both sides.

“People hire people,” Duke said. AI can help in the process, analyzing new groups of candidates in a world where a billion people can see your job advert in an instant. “Where it works is when you combine the best technology with the best exceptional human recruiting skills,” Duke said. It is not human ‘in the loop’ as human ‘in the lead’. Soft skills are important, just as much as the tech engineer as the executive lost when the chat at the water cooler became AI.