Throughout our conversations with employers in the past year, a new question continues to come up as AI reshapes the workplace: how do you know whether someone can exercise their judgment skills when the work around them is changing at a rapid pace? In our brief with Business Roundtable, Assessing Durable Skills in the Age of AI, talent leaders told us the strongest evidence of skills like judgment comes from watching someone demonstrate those skills to achieve a desired outcome. Many of the same talent leaders also reported that intentional development of these skills is sometimes concentrated in leadership programs, even though employees at every level stand to benefit greatly from it. At America Succeeds, we’re increasing our focus on identifying the skill signals employers trust and how internal development impacts business results. Two recent reports speak to why these issues are front of mind for employers.
In a new McKinsey article titled The Judgment Bottleneck: Balancing AI and human speeds in strategy, the authors share an illustrative example from a business leader they interviewed. He used to ask junior analysts basic questions about how their models are built, such as whether a forecast used nominal GDP. “Now people don’t know,” he says. The work of building the models that would once have taught them the intricacies of how they operate is now being automated. Using AI to automate and speed up technical tasks provides companies short-term value, but that newfound efficiency simultaneously stifles opportunities to develop the knowledge upon which skills like judgment are built. The authors put it this way: “As analysis accelerates, judgment becomes the new bottleneck.”
Marsh’s The Agility Gap, written by Tara Cooper in collaboration with Guild, reaches a similar conclusion from the HR side. As AI takes on more procedural and technical work, the report argues, “human judgment becomes more valuable,” and building it “requires deliberate learning experiences most teams are not currently funding.” Both reports point to the same structural change in the workplace. Guild notes that entry-level and mid-level roles, which have long been the incubator for skill development, are being automated or hollowed out. McKinsey describes the outcome of this effect: “As the volume of output grows, less experienced employees may increasingly be asked to make decisions for which they lack expertise.”
Employers are working on this, though progress is uneven. Marsh’s AI Readiness Report found that 54% of organizations feel prepared to reskill their workforce as AI changes roles. Employees, meanwhile, are signing up for AI training in large numbers: across Guild’s learners, enrollment in AI and machine learning programs has grown tenfold over the past two years. However, The Agility Gap suggests that training on durable skills like judgment draws less attention on its own. One head of L&D at a large bank works around this by building lessons on judgment and ethics into her AI courses: “If I go out with an AI training, everybody signs up… But inside that training, I’m spending a lot of time talking about good judgment, governance, ethics.” Whether that training sticks depends on practice: McKinsey’s authors write that people build judgment “through active practice, feedback, repetition, and low-stakes error correction.”
For HR and L&D leaders, the practical step is building judgment into the work AI is already changing. At Rocket, Chief Talent Officer Dr. KimArie Yowell “puts leaders at the center of the AI accountability layer.” They are expected to tell when someone hands over a raw AI output instead of work they shaped themselves, and to press their teams on the decisions behind it. Rocket also holds show-and-tells to give team members space to present real problems they solved. This approach allows entry- and mid-level talent to practice explaining and defending their own judgment calls. Rocket and other like-minded companies are acknowledging that it matters who gets the durable skills practice. When judgment-building exercises are reserved mainly for leadership programs, employees in early-career roles – where AI is absorbing many of the tasks that once built knowledge and decision-making – get the fewest chances to develop it. AdventHealth offers another model to address this. Marquita Williams, who leads organizational learning and development, replaced limited licenses with on-demand learning for every team member. She built a three-year plan around durable skills, starting with emotional intelligence and communication, because “those durable soft skills will have the longest shelf life.”
Knowing how well any of this works requires better evidence than course completions. Guild recommends tracking internal redeployment rates alongside upskilling data. Our work with Business Roundtable found that relatively few companies have built systems that assess durable skills repeatedly against a consistent standard. Without that data, it is challenging to identify the ROI of durable skills training and connect development spending to internal redeployment. That is where our research is headed, and we want to learn from the employers already working on it.
If you are a talent or learning leader, ask what evidence your development programs give you that employees can apply judgment in their roles. Assessing Durable Skills in the Age of AI lays out four properties of a strong skill signal and gives you a framework to check them against.
Already building this kind of system? We’re studying how employers read and weigh evidence of durable skills, and we want to learn from the companies already doing the work. Get in touch.
