The bottom rung returns as AI reshapes entry-level jobs

For months, the narrative around AI and work has centered on loss—especially for those just starting out. But early data and hiring plans suggest a more complicated story. IBM, for one, is expanding entry-level hiring and redefining entry-level roles for an AI-first workplace, and new data from advisory firm Teneo indicates the company may not be in the minority: 67% of global CEOs say AI is increasing entry-level headcount, not reducing it.

“Entry-level roles are shifting from purely task-driven work to analysis, problem-solving and responsible AI use,” said Natasha Pillay-Bemath, IBM’s VP of Global Talent Acquisition and Executive Search, to IBM Think. She noted that learning agility now matters as much as technical skills for new hires.

Meanwhile, the broader picture may be less apocalyptic than the headlines suggest. While fears about AI-driven job losses persist, The Budget Lab at Yale recently found that concerns about AI displacing today’s workforce—including entry-level workers—remain “largely speculative,” suggesting that it’s too soon to know how disruptive it will be.

And despite chatter about shrinking tech roles, Citadel Securities reports that software engineering job postings are up 11% year over year, a sign that demand for technical talent is holding steady, at least for now.

Against this backdrop, IBM is moving early and plans to triple US entry-level hiring in 2026, said IBM Chief Human Resources Officer Nickle LaMoreaux at a recent Charter AI Summit in New York. LaMoreaux said that investing in entry-level talent is essential to IBM’s long-term agility. “If we don’t continue to invest in entry-level hires, what happens in 3–5 years?” she asked. “There’s no pipeline; the well simply dries up.”

IBM is hiring entry-level talent across all business units including software, consulting, infrastructure and marketing, said Pillay-Bemath. The job roles for this workforce also run the gamut from developers to quantum data scientists to social media and influencer marketers. The reality, LaMoreaux said, is that enterprises must “rewrite every job” in this AI era.

This shift mirrors a growing group of companies pushing back on the entry-level hire doomsayers. McKinsey is planning a 12% increase in North American hiring in 2026 and emphasizes that entry-level roles are evolving, not disappearing. As problem-solving skills become increasingly critical on day one, McKinsey now uses a gamified assessment tool called Solve to screen new applicants, said Heather Stefanski, Chief Learning and Development Officer at McKinsey, at the Charter conference in New York. In this virtual ecosystem, applicants are tested on critical thinking, decision-making and systems thinking—rather than prior business knowledge.

“There is a polarity,” said Stefanski. “We are doubling down on what makes you uniquely human—and inserting more tech.” IT consulting and outsourcing firm Cognizant is also expanding its entry-level recruitment, including hiring more liberal arts and non-STEM graduates. “AI is an amplifier of human potential,” said Cognizant CEO Ravi Kumar in a recent interview in Fortune. “It’s not a displacement strategy.”

Beyond hiring increases, what does redesigning these jobs actually look like? Entry-level roles are starting to prioritize systems thinking, critical analysis and human oversight of AI, rather than repetitive execution. “As AI handles more routine coding and documentation, [entry-level] professionals are increasingly expected to think holistically—understanding systems end-to-end and validating AI outputs for quality and bias,” said Pillay-Bemath. Entry-level developers are now working with real clients far earlier in their careers, she said.

But while AI can automate workflows and speed up the development of new talent, “it can’t understand a team’s broader goals, navigate ambiguity or bring human judgment to complex decisions,” said Pillay-Bemath. “It also can’t build trust or apply contextual judgement in client operations.” As a result, enterprises need to double down on training to develop these human skills that AI cannot automate.

Reshaping jobs, and particularly entry-level jobs, is what motivated Matt Beane, Associate Professor of Technology Management at UC Santa Barbara, to create a new company called SkillBench. In an interview with IBM Think, he said he wanted to help companies develop a structured framework, so that instead of a “give everyone AI and good luck” approach, teams could “steer into the skid” by restructuring tasks and reviews.

One training model he currently recommends to software clients consists of small apprentice teams, led by a senior employee who coaches a small cohort of newbies toward production-ready outcomes. During a compressed timeframe of a few days, the coach challenges the them to use AI tools to develop and test out approaches to “an utterly inappropriately difficult goal such as ‘I want you to create an entire backend that can handle a million users with a million changes to the database every minute.”

Then, the coach deploys a “Socratic critique”—that is, he asks students questions to understand their process, find alternate solutions and uncover assumptions. This relationship is also bi-directional, so that the newer employees expose senior workers to new tools. Recently, Beane has also started meeting with law, finance and accounting firms. Since much of their work is “rigorously verifiable … quite codelike,” Beane said, teams in these knowledge-work fields can borrow this technique from software talent training.

Another key element to the SkillBench approach is mapping out a company’s actual work in granular detail. That way, Beane said, leaders can identify which tasks offer meaningful skill development, rather than making decisions based on “vibes and anecdotes.” Beane works with companies to map team workflows into “tiny packets,” then scores each task along two axes: “where AI gives a big productivity boost, and where the task offers meaningful skill development.” Then, he uses that 2×2 grid to protect high-learning tasks from over-automation while greenlighting automation in “skill deserts,” work that doesn’t develop any skills.

Finally, he pairs this organizational dashboard with a private dashboard for individual employees—a “Strava for AI,” as he calls it—that tracks chat logs and code differences for example. In this way, workers can see personal progress using AI tools while leaders see only group-level best practices.

Looking forward, IBM experts stressed that mindset will be much more important than technical capabilities. “That’s exactly what entry-level talent brings: curiosity, adaptability and a commitment to continuous learning,” IBM’s Pillay-Bemath said. “AI can boost productivity, but it can’t develop the next generation of technical leaders or innovators.” LaMoreaux added that the companies that will be most successful in the next few years are “investing in talent now to set themselves up for the future.”