{"id":326041,"date":"2026-09-19T05:00:00","date_gmt":"2026-09-19T02:00:00","guid":{"rendered":"https:\/\/ceotudent.com\/t-shaped-professional-obsolete-new-shape-ai-era-careers"},"modified":"2026-09-19T05:00:00","modified_gmt":"2026-09-19T02:00:00","slug":"t-shaped-professional-obsolete-new-shape-ai-era-careers","status":"publish","type":"post","link":"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers","title":{"rendered":"The T-Shaped Professional Is Obsolete: The New Shape for AI-Era Careers"},"content":{"rendered":"<p><strong>TL;DR.<\/strong> The T-shaped professional, deep in one discipline and broad enough to collaborate across others, rested on two prices: depth was scarce, and breadth was expensive to acquire. Generative AI has changed both. In a field experiment with Procter &amp; Gamble professionals, individuals working with AI improved solution quality by 0.37 standard deviations, matching two-person teams working without AI, and produced balanced technical and commercial solutions regardless of their own background. The horizontal bar of the T can now be borrowed. The vertical bar is changing too: in a published study of 5,172 customer-support agents, AI assistance raised productivity by 15% on average, with the least experienced workers improving most. Our own analysis joins O*NET 31.0 skill ratings with published AI-exposure scores for 372 US occupations that typically require a degree. Occupations high in analytic skills but low in people skills have the highest average exposure (0.52); pairing the same analytic intensity with strong people skills is associated with 16% lower exposure. The new shape we propose is the Tree: a canopy of AI-extended breadth you borrow, a trunk of judgment depth you own, roots of relationships and context that machines cannot supply, and growth rings that record how fast you keep learning.<\/p>\n<p>This piece is part of our series on career strategy in the AI era. It builds on <a href=\"https:\/\/ceotudent.com\/en\/from-specialist-to-orchestrator-career-transition-framework-ai-era\">from specialist to orchestrator<\/a>, <a href=\"https:\/\/ceotudent.com\/en\/skill-stacking-ai-era\">skill stacking in the AI era<\/a> and <a href=\"https:\/\/ceotudent.com\/en\/career-capital-ai-era-what-compounds-what-decays\">career capital: what compounds and what decays<\/a>.<\/p>\n<div id=\"ez-toc-container\" class=\"ez-toc-v2_0_84 counter-hierarchy ez-toc-counter ez-toc-grey ez-toc-container-direction\">\n<div class=\"ez-toc-title-container\">\n<p class=\"ez-toc-title\" style=\"cursor:inherit\">Table of Contents<\/p>\n<span class=\"ez-toc-title-toggle\"><a href=\"#\" class=\"ez-toc-pull-right ez-toc-btn ez-toc-btn-xs ez-toc-btn-default ez-toc-toggle\" aria-label=\"Toggle Table of Content\"><span class=\"ez-toc-js-icon-con\"><span class=\"\"><span class=\"eztoc-hide\" style=\"display:none;\">Toggle<\/span><span class=\"ez-toc-icon-toggle-span\"><svg style=\"fill: #999;color:#999\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" class=\"list-377408\" width=\"20px\" height=\"20px\" viewBox=\"0 0 24 24\" fill=\"none\"><path d=\"M6 6H4v2h2V6zm14 0H8v2h12V6zM4 11h2v2H4v-2zm16 0H8v2h12v-2zM4 16h2v2H4v-2zm16 0H8v2h12v-2z\" fill=\"currentColor\"><\/path><\/svg><svg style=\"fill: #999;color:#999\" class=\"arrow-unsorted-368013\" xmlns=\"http:\/\/www.w3.org\/2000\/svg\" width=\"10px\" height=\"10px\" viewBox=\"0 0 24 24\" version=\"1.2\" baseProfile=\"tiny\"><path d=\"M18.2 9.3l-6.2-6.3-6.2 6.3c-.2.2-.3.4-.3.7s.1.5.3.7c.2.2.4.3.7.3h11c.3 0 .5-.1.7-.3.2-.2.3-.5.3-.7s-.1-.5-.3-.7zM5.8 14.7l6.2 6.3 6.2-6.3c.2-.2.3-.5.3-.7s-.1-.5-.3-.7c-.2-.2-.4-.3-.7-.3h-11c-.3 0-.5.1-.7.3-.2.2-.3.5-.3.7s.1.5.3.7z\"\/><\/svg><\/span><\/span><\/span><\/a><\/span><\/div>\n<nav><ul class='ez-toc-list ez-toc-list-level-1 ' ><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-1\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#Where-the-T-came-from-and-the-two-bets-it-made\" >Where the T came from, and the two bets it made<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-2\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#What-AI-does-to-the-horizontal-bar\" >What AI does to the horizontal bar<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-3\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#What-AI-does-to-the-vertical-bar\" >What AI does to the vertical bar<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-4\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#Our-analysis-which-skill-profiles-are-most-exposed\" >Our analysis: which skill profiles are most exposed<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-5\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#Why-the-T-is-obsolete-not-just-outdated\" >Why the T is obsolete, not just outdated<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-6\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#The-new-shape-the-Tree\" >The new shape: the Tree<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-7\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#The-early-career-problem-growing-a-trunk-when-the-first-rungs-thin-out\" >The early-career problem: growing a trunk when the first rungs thin out<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-8\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#How-to-audit-your-own-shape\" >How to audit your own shape<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-9\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#When-the-T-is-still-the-right-shape\" >When the T is still the right shape<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-10\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#Frequently-asked-questions\" >Frequently asked questions<\/a><\/li><li class='ez-toc-page-1 ez-toc-heading-level-2'><a class=\"ez-toc-link ez-toc-heading-11\" href=\"https:\/\/ceotudent.com\/en\/t-shaped-professional-obsolete-new-shape-ai-era-careers\/#Sources\" >Sources<\/a><\/li><\/ul><\/nav><\/div>\n<h2 id=\"where-the-t-came-from-and-the-two-bets-it-made\"><span class=\"ez-toc-section\" id=\"Where-the-T-came-from-and-the-two-bets-it-made\"><\/span>Where the T came from, and the two bets it made<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The T is older than most people who use it. The earliest use that researchers usually cite is a short 1991 article by David Guest in The Independent, headlined &ldquo;The hunt is on for the Renaissance Man of computing&rdquo;. Harvard Business Review brought the idea to management audiences in 2001 with Morten Hansen and Bettina von Oetinger&rsquo;s &ldquo;Introducing T-shaped managers&rdquo;. IDEO&rsquo;s chief executive Tim Brown gave the definition most people now repeat in a 2010 interview with Chief Executive magazine: the vertical stroke is &ldquo;a depth of skill that allows them to contribute to the creative process&rdquo;, and the horizontal stroke is the disposition for collaboration across disciplines.<\/p>\n<p>Behind the letter sat two economic bets:<\/p>\n<ol>\n<li><strong>Depth is the scarce asset.<\/strong> You are hired for the vertical bar. Years of study and practice in one field are hard to copy, so they earn a premium.<\/li>\n<li><strong>Breadth is expensive, so a little is enough.<\/strong> You do not need to know marketing if you are an engineer; you need enough of it to talk to the marketer. The horizontal bar is a collaboration interface, not a second skill.<\/li>\n<\/ol>\n<p>Both bets were reasonable for thirty years. Generative AI changes the price of each.<\/p>\n<h2 id=\"what-ai-does-to-the-horizontal-bar\"><span class=\"ez-toc-section\" id=\"What-AI-does-to-the-horizontal-bar\"><\/span>What AI does to the horizontal bar<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The clearest evidence comes from &ldquo;The Cybernetic Teammate&rdquo;, a preregistered field experiment at Procter &amp; Gamble by Fabrizio Dell&rsquo;Acqua, Karim Lakhani and colleagues, published in Organization Science in 2026. Professionals worked on real product-innovation challenges, alone or in two-person teams of one R&amp;D and one commercial professional, with or without a GPT-4-based tool.<\/p>\n<p>In the working-paper version, teams without AI improved solution quality by 0.24 standard deviations over individuals working alone. Individuals with AI improved by 0.37 standard deviations, and teams with AI by 0.39. The published abstract summarises it plainly: individuals with AI matched the performance of teams without AI.<\/p>\n<p>The finding that matters most for the T is about silos. Without AI, R&amp;D professionals tended to propose technical solutions and commercial professionals commercial ones. In the authors&rsquo; words, professionals using AI produced more balanced solutions, &ldquo;regardless of their professional background&rdquo;. That is the horizontal bar, delivered on demand.<\/p>\n<p>There is an important detail in the same study, and it matters later. The highest-quality ideas still came from people working together: teams with AI were 9.2 percentage points more likely than the control group (mean 5.8%) to produce a solution in the top 10%, while the effect for individuals with AI was not statistically significant. AI supplies breadth of knowledge. It did not replace the value of another human in the room.<\/p>\n<h2 id=\"what-ai-does-to-the-vertical-bar\"><span class=\"ez-toc-section\" id=\"What-AI-does-to-the-vertical-bar\"><\/span>What AI does to the vertical bar<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Depth is being repriced as well, but unevenly.<\/p>\n<p>In &ldquo;Generative AI at Work&rdquo;, published in the Quarterly Journal of Economics in 2025, Erik Brynjolfsson, Danielle Li and Lindsey Raymond studied 5,172 customer-support agents given an AI assistant. Productivity, measured as issues resolved per hour, rose 15% on average. Less experienced and lower-skilled workers improved both speed and quality, while the most experienced and highest-skilled workers saw small gains in speed and small declines in quality. Know-how that took years to build became partly available to newcomers through the tool.<\/p>\n<p>At the same time, depth is what tells you when the tool is wrong. In the Boston Consulting Group experiment with 758 consultants, published in Organization Science in 2026 (&ldquo;Navigating the Jagged Technological Frontier&rdquo;), those given GPT-4 completed 12.2% more tasks and worked 25.1% faster on tasks inside the AI frontier. On a task deliberately chosen to sit outside it, consultants without AI were correct about 84.5% of the time, while those with AI scored 60% and 70.6%, an average drop of 19 percentage points. Using AI on a task beyond its capabilities made people less accurate, and depth in the domain is what lets you notice that you are on such a task.<\/p>\n<p>The original GPTs-are-GPTs study by Tyna Eloundou, Sam Manning, Pamela Mishkin and Daniel Rock points in the same direction at the level of skills: across occupations, the importance of programming and writing skills was strongly positively associated with exposure to large language models, while science and critical thinking skills were strongly negatively associated.<\/p>\n<p>So the vertical bar is splitting in two. <strong>Codified depth<\/strong>, the part of expertise that can be written down and pattern-matched, is getting cheaper. <strong>Judgment depth<\/strong>, the part that lets you verify, decide and take responsibility, is becoming the reason anyone needs you.<\/p>\n<h2 id=\"our-analysis-which-skill-profiles-are-most-exposed\"><span class=\"ez-toc-section\" id=\"Our-analysis-which-skill-profiles-are-most-exposed\"><\/span>Our analysis: which skill profiles are most exposed<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>To test this on current data, we joined two public datasets that are not usually combined:<\/p>\n<ul>\n<li><strong>O*NET 31.0<\/strong> (US Department of Labor, August 2026): importance ratings, on a 1-to-5 scale, for 35 skills in each of 910 occupations.<\/li>\n<li><strong>Occupation-level exposure scores<\/strong> published by Eloundou and colleagues with their study. We used their beta measure: the share of an occupation&rsquo;s tasks where access to a large language model, directly or through software built on it, could cut the time needed by at least half (the second route is weighted at 0.5). The dataset has two versions of the score, one from human annotators and one from GPT-4.<\/li>\n<\/ul>\n<p>Across all 910 occupations, almost every cognitive skill is positively correlated with exposure, because exposure mostly separates desk work from physical work. That comparison says little about a knowledge worker&rsquo;s choices. So we narrowed the analysis to the 372 occupations in O*NET Job Zones 4 and 5, where most jobs require a bachelor&rsquo;s degree or a graduate degree. Within that group, the picture splits.<\/p>\n<p><strong>Table 1. How skill importance relates to AI exposure in degree-level occupations (CEOtudent calculation, O*NET 31.0 and Eloundou et al. exposure data)<\/strong><\/p>\n<table>\n<thead>\n<tr>\n<th>Skill<\/th>\n<th>Correlation with exposure, human-rated<\/th>\n<th>Correlation with exposure, GPT-4-rated<\/th>\n<th>For comparison: all 910 occupations, human-rated<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Programming<\/td>\n<td>+0.35<\/td>\n<td>+0.46<\/td>\n<td>+0.57<\/td>\n<\/tr>\n<tr>\n<td>Systems analysis<\/td>\n<td>+0.21<\/td>\n<td>+0.13<\/td>\n<td>+0.58<\/td>\n<\/tr>\n<tr>\n<td>Mathematics<\/td>\n<td>+0.19<\/td>\n<td>+0.21<\/td>\n<td>+0.45<\/td>\n<\/tr>\n<tr>\n<td>Writing<\/td>\n<td>+0.12<\/td>\n<td>0.00<\/td>\n<td>+0.74<\/td>\n<\/tr>\n<tr>\n<td>Critical thinking<\/td>\n<td>-0.07<\/td>\n<td>-0.13<\/td>\n<td>+0.59<\/td>\n<\/tr>\n<tr>\n<td>Judgment and decision making<\/td>\n<td>-0.12<\/td>\n<td>-0.21<\/td>\n<td>+0.55<\/td>\n<\/tr>\n<tr>\n<td>Science<\/td>\n<td>-0.17<\/td>\n<td>-0.20<\/td>\n<td>+0.20<\/td>\n<\/tr>\n<tr>\n<td>Instructing<\/td>\n<td>-0.23<\/td>\n<td>-0.39<\/td>\n<td>+0.45<\/td>\n<\/tr>\n<tr>\n<td>Coordination<\/td>\n<td>-0.28<\/td>\n<td>-0.37<\/td>\n<td>+0.36<\/td>\n<\/tr>\n<tr>\n<td>Service orientation<\/td>\n<td>-0.35<\/td>\n<td>-0.48<\/td>\n<td>+0.41<\/td>\n<\/tr>\n<tr>\n<td>Social perceptiveness<\/td>\n<td>-0.36<\/td>\n<td>-0.51<\/td>\n<td>+0.48<\/td>\n<\/tr>\n<tr>\n<td>Monitoring<\/td>\n<td>-0.36<\/td>\n<td>-0.46<\/td>\n<td>+0.31<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Spearman rank correlations across 372 occupations in Job Zones 4 and 5, except the last column. Positive values mean occupations that rate the skill as more important tend to have higher exposure.<\/p>\n<p>Two patterns stand out. Among degree-level jobs, the analytic production skills that usually form the vertical bar of a technical T, programming, systems analysis and mathematics, go with <strong>higher<\/strong> exposure. The people skills that the T treated as a thin collaboration interface, social perceptiveness, coordination, instructing and service orientation, go with <strong>lower<\/strong> exposure. The same direction appears in both the human-rated and the GPT-4-rated scores.<\/p>\n<p>To see what this means for a profile rather than a single skill, we built two simple composites. An analytic score averages programming, systems analysis, systems evaluation, mathematics and operations analysis. A people score averages social perceptiveness, coordination, instructing, service orientation, persuasion and negotiation. We split the 372 occupations at the median of each.<\/p>\n<p><strong>Table 2. Average AI exposure by skill profile, degree-level occupations (CEOtudent calculation)<\/strong><\/p>\n<table>\n<thead>\n<tr>\n<th>Profile<\/th>\n<th>Occupations<\/th>\n<th>Mean exposure, human-rated<\/th>\n<th>Mean exposure, GPT-4-rated<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>High analytic, low people (the classic technical T)<\/td>\n<td>123<\/td>\n<td>0.52<\/td>\n<td>0.57<\/td>\n<\/tr>\n<tr>\n<td>High analytic, high people<\/td>\n<td>63<\/td>\n<td>0.43<\/td>\n<td>0.46<\/td>\n<\/tr>\n<tr>\n<td>Low analytic, high people<\/td>\n<td>123<\/td>\n<td>0.41<\/td>\n<td>0.42<\/td>\n<\/tr>\n<tr>\n<td>Low analytic, low people<\/td>\n<td>63<\/td>\n<td>0.44<\/td>\n<td>0.51<\/td>\n<\/tr>\n<tr>\n<td>All 372 degree-level occupations<\/td>\n<td>372<\/td>\n<td>0.45<\/td>\n<td>0.49<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>The classic technical T, deep analytic skill with a thin people layer, is the most exposed profile in the group. Occupations that combine the same analytic intensity with strong people skills average 0.43 instead of 0.52 on the human-rated score, 16% lower, and 19% lower on the GPT-4-rated score. Ranked by the people composite alone, the top quarter of degree-level occupations averages 0.40 against 0.50 for the bottom quarter.<\/p>\n<p>Four cautions before you act on this. Exposure is not job loss: it measures where AI could speed up tasks, and the Canaries study below shows that the labour-market effect depends on whether AI substitutes for workers or complements them. The exposure scores were produced in 2023 with GPT-4-era capabilities. These are occupation averages, and a person can have a very different profile from the typical holder of their job title. And correlation is not causation: people-heavy occupations may be less exposed for reasons other than the people skills themselves. Even so, the direction is consistent with the P&amp;G and BCG experiments, and with David Deming&rsquo;s 2017 finding in the Quarterly Journal of Economics that between 1980 and 2012 jobs requiring high levels of social interaction grew by nearly 12 percentage points as a share of the US labour force, while math-intensive but less social jobs shrank by 3.3 percentage points, with the strongest growth in jobs needing both.<\/p>\n<h2 id=\"why-the-t-is-obsolete-not-just-outdated\"><span class=\"ez-toc-section\" id=\"Why-the-T-is-obsolete-not-just-outdated\"><\/span>Why the T is obsolete, not just outdated<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>A letter shape is a model, and a model is obsolete when its assumptions no longer describe the world. Both bars of the T hid two different things under one stroke.<\/p>\n<ul>\n<li>The <strong>horizontal bar<\/strong> mixed breadth of <em>knowledge<\/em> (knowing enough marketing, finance or law to follow the conversation) with breadth of <em>relationships<\/em> (being trusted by, and able to coordinate with, people in those functions). AI now supplies much of the first. It does not supply the second.<\/li>\n<li>The <strong>vertical bar<\/strong> mixed <em>codified<\/em> depth (procedures, syntax, standard analyses) with <em>judgment<\/em> depth (knowing what good looks like, spotting the confident error, owning the decision). AI compresses the first. It raises the value of the second.<\/li>\n<\/ul>\n<p>A shape that cannot tell these apart gives bad advice. It tells an analyst to go deeper into the most exposed part of their skill set, and it tells everyone that a thin layer of collaboration skill is enough.<\/p>\n<h2 id=\"the-new-shape-the-tree\"><span class=\"ez-toc-section\" id=\"The-new-shape-the-Tree\"><\/span>The new shape: the Tree<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The framework below is our editorial synthesis. We call it the Tree, partly because a T with the right additions becomes one, and partly because trees grow every year.<\/p>\n<p><strong>Table 3. From the T to the Tree (CEOtudent editorial framework)<\/strong><\/p>\n<table>\n<thead>\n<tr>\n<th>Part of the Tree<\/th>\n<th>What it is<\/th>\n<th>What the T assumed<\/th>\n<th>What changed, and the evidence<\/th>\n<th>What to build<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Canopy<\/td>\n<td>Breadth of knowledge across functions, extended by AI<\/td>\n<td>Breadth is expensive; a thin bar is enough<\/td>\n<td>AI gives individuals cross-functional breadth; P&amp;G professionals using AI produced balanced solutions regardless of background<\/td>\n<td>Fluency in using AI to enter adjacent domains fast, plus enough literacy to ask good questions there<\/td>\n<\/tr>\n<tr>\n<td>Trunk<\/td>\n<td>Judgment depth in one domain: the ability to verify, decide and be accountable<\/td>\n<td>Depth means codified expertise<\/td>\n<td>Codified know-how diffuses to novices (15% average gain, largest for less experienced agents); outside the AI frontier, users without depth were 19 points less likely to be right<\/td>\n<td>Depth aimed at evaluation: knowing failure modes, standards of quality and where AI tends to be wrong in your field<\/td>\n<\/tr>\n<tr>\n<td>Roots<\/td>\n<td>Relationships, trust and context: coordination, social perceptiveness, instructing, knowledge of how your organisation and market actually work<\/td>\n<td>Collaboration is an interface skill<\/td>\n<td>In our analysis, people skills go with lower exposure among degree-level jobs; top-10% ideas at P&amp;G came from teams, not individuals with AI<\/td>\n<td>Deliberate investment in the people and settings where your work lands<\/td>\n<\/tr>\n<tr>\n<td>Growth rings<\/td>\n<td>Learning rate: how quickly you rebuild each part as tools change<\/td>\n<td>Skills, once built, last a career<\/td>\n<td>Employers expect two-fifths (39%) of workers&rsquo; existing skill sets to be transformed or become outdated between 2025 and 2030 (WEF)<\/td>\n<td>A renewal cadence: a scheduled review of what you know, what decayed and what to relearn<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Three differences from the T are worth making explicit.<\/p>\n<p><strong>The canopy is borrowed, and that is fine.<\/strong> You do not need to own the knowledge in the canopy, but you need to know how to reach it and judge it. A wide canopy with no trunk is exactly the profile the BCG experiment warned about: confident output in areas where you cannot tell right from wrong.<\/p>\n<p><strong>The trunk is about evaluation, not production.<\/strong> The question is no longer &ldquo;can you do the work faster than anyone?&rdquo; but &ldquo;can you tell whether the work is right, and will you put your name on it?&rdquo; That is also the capability at the centre of our <a href=\"https:\/\/ceotudent.com\/en\/from-specialist-to-orchestrator-career-transition-framework-ai-era\">orchestrator framework<\/a> and of <a href=\"https:\/\/ceotudent.com\/en\/discernment-gap-telling-ai-output-from-human-thinking\">the discernment gap<\/a>.<\/p>\n<p><strong>The roots are invisible, which is why the T ignored them.<\/strong> Relationships, trust and organisational context do not appear on a skills list, but they decide whether good analysis turns into action. They are also the part of a career that is hardest to copy with a subscription.<\/p>\n<h2 id=\"the-early-career-problem-growing-a-trunk-when-the-first-rungs-thin-out\"><span class=\"ez-toc-section\" id=\"The-early-career-problem-growing-a-trunk-when-the-first-rungs-thin-out\"><\/span>The early-career problem: growing a trunk when the first rungs thin out<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The Tree raises a hard question for people starting out. Judgment depth has traditionally been built by doing years of codified work, the same work AI now does quickly. If that work disappears, how does anyone grow a trunk?<\/p>\n<p>The early data is a warning. In the August 2026 version of &ldquo;Canaries in the Coal Mine?&rdquo;, Erik Brynjolfsson, Bharat Chandar and Ruyu Chen used ADP payroll data covering millions of US workers through June 2026. They found no evidence of widespread, economy-wide job displacement, but employment of workers aged 22 to 25 in AI-exposed occupations stood 19% below where it would have been had it kept pace with less-exposed peers, with no comparable gap for experienced workers. The declines were concentrated where AI usage primarily substitutes for human tasks; where usage primarily complements workers, employment was flat or rising. The authors describe these as early, descriptive indicators rather than causal estimates.<\/p>\n<p>A Danish study points the other way on the size of the effect so far. Anders Humlum and Emilie Vestergaard, in NBER Working Paper 33777 (revised March 2026), used administrative records and found precise null effects on earnings and recorded hours, ruling out effects larger than 2% two years after the launch of ChatGPT. What changed, they report, was the structure of work: employers absorbed AI by reorganising tasks.<\/p>\n<p>Read together: the aggregate shock is still small, but the ladder is changing shape at the bottom. For early-career readers, that means deliberately seeking work that builds judgment, not just output: reviewing, deciding, explaining decisions to others, handling the exceptions the AI gets wrong. For managers, it means designing roles where juniors still get to make judgment calls under supervision, or the organisation will have no trunks in ten years.<\/p>\n<h2 id=\"how-to-audit-your-own-shape\"><span class=\"ez-toc-section\" id=\"How-to-audit-your-own-shape\"><\/span>How to audit your own shape<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The audit below turns the Tree into questions. Score each from 1 (not true) to 5 (clearly true). It is a self-reflection tool, not a validated instrument.<\/p>\n<p><strong>Table 4. The Tree audit (CEOtudent editorial framework)<\/strong><\/p>\n<table>\n<thead>\n<tr>\n<th>Part<\/th>\n<th>Question<\/th>\n<th>Warning sign if you score 1 or 2<\/th>\n<th>First move<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Canopy<\/td>\n<td>Can I use AI to get a working understanding of an adjacent function within a week?<\/td>\n<td>You avoid problems outside your specialty<\/td>\n<td>Pick one adjacent function and run a two-week AI-assisted learning sprint with real questions from your work<\/td>\n<\/tr>\n<tr>\n<td>Canopy<\/td>\n<td>Do I know when my borrowed knowledge is too thin to act on?<\/td>\n<td>You repeat AI summaries you cannot defend<\/td>\n<td>Before acting outside your field, ask one expert to check the AI&rsquo;s answer<\/td>\n<\/tr>\n<tr>\n<td>Trunk<\/td>\n<td>Can I reliably spot when AI output in my field is wrong?<\/td>\n<td>You check AI output only for typos and tone<\/td>\n<td>Keep a log of AI errors you catch; review it monthly<\/td>\n<\/tr>\n<tr>\n<td>Trunk<\/td>\n<td>Do I own decisions, not just deliverables?<\/td>\n<td>Your work ends when you hand it over<\/td>\n<td>Write a short decision memo for your next important recommendation<\/td>\n<\/tr>\n<tr>\n<td>Roots<\/td>\n<td>Would people in two other functions call me first about a problem?<\/td>\n<td>Your network is all in your own discipline<\/td>\n<td>Schedule one working session a month with a different function<\/td>\n<\/tr>\n<tr>\n<td>Roots<\/td>\n<td>Do I understand how decisions really get made where my work lands?<\/td>\n<td>Good analysis of yours often goes unused<\/td>\n<td>Map who decides, who influences and who implements for one current project<\/td>\n<\/tr>\n<tr>\n<td>Rings<\/td>\n<td>Have I rebuilt a significant part of my skill set in the past 12 months?<\/td>\n<td>Your methods look the same as three years ago<\/td>\n<td>Set a quarterly review: what I learned, what decayed, what to relearn<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>If most of your low scores are in the trunk, deepen your ability to evaluate before you broaden. If they are in the roots, your analytic strength is probably sitting in the most exposed quadrant of Table 2. If they are in the canopy, you are underusing the cheapest resource you have. Our guide to <a href=\"https:\/\/ceotudent.com\/en\/how-to-develop-good-judgment-expert-intuition-research\">developing good judgment<\/a> goes deeper on the trunk.<\/p>\n<p>This is the CEO and student lens in one shape. The CEO half owns the trunk and tends the roots: decisions, accountability, relationships. The student half keeps the canopy growing and adds a ring every year.<\/p>\n<h2 id=\"when-the-t-is-still-the-right-shape\"><span class=\"ez-toc-section\" id=\"When-the-T-is-still-the-right-shape\"><\/span>When the T is still the right shape<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>The Tree is a default for knowledge work, not a law.<\/p>\n<ul>\n<li><strong>Where AI use is restricted.<\/strong> In settings where regulation, security or confidentiality limits AI tools, breadth is still expensive and the classic T still pays.<\/li>\n<li><strong>In deep technical frontier work.<\/strong> Researchers pushing a field forward still need extreme codified depth; the AI frontier is often behind them.<\/li>\n<li><strong>Where the scarce thing really is execution.<\/strong> Some roles are valued for production speed under tight standards, and a narrow, deep profile fits them.<\/li>\n<li><strong>Early in a career, for a while.<\/strong> Some codified depth is how judgment gets built. The goal is not to skip it but to not stop there.<\/li>\n<\/ul>\n<p>The honest summary of the evidence is that the shape is changing faster than the labour market, not that careers built on the T are collapsing. That is a reason to start growing the Tree now, while the cost of changing is low.<\/p>\n<h2 id=\"frequently-asked-questions\"><span class=\"ez-toc-section\" id=\"Frequently-asked-questions\"><\/span>Frequently asked questions<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p><strong>What is a T-shaped professional?<\/strong><br \/>\nA T-shaped professional has deep expertise in one discipline (the vertical stroke) and a broader ability to collaborate across other disciplines (the horizontal stroke). The term appeared in print at least as early as 1991 and was popularised in management by Harvard Business Review in 2001 and by IDEO&rsquo;s Tim Brown.<\/p>\n<p><strong>Is the T-shaped model still relevant in the AI era?<\/strong><br \/>\nPartly. Its core idea, combining depth with breadth, still holds, but its assumptions do not. AI now supplies much of the breadth of knowledge the horizontal bar represented, and it compresses codified depth. What stays scarce is judgment depth and the relationships and context that turn analysis into action.<\/p>\n<p><strong>What is the Tree-shaped professional?<\/strong><br \/>\nIt is our framework for AI-era careers: a canopy of AI-extended breadth you can borrow, a trunk of judgment depth you own, roots of relationships and organisational context, and growth rings for your rate of learning.<\/p>\n<p><strong>Which skills are least exposed to AI in professional jobs?<\/strong><br \/>\nIn our analysis of 372 US occupations that typically require a degree, occupations that rate social perceptiveness, monitoring, service orientation, coordination and instructing as important tended to have lower AI exposure, while programming, systems analysis and mathematics went with higher exposure.<\/p>\n<p><strong>Should I stop going deeper in my specialty?<\/strong><br \/>\nNo. Aim your depth at evaluation rather than production: the ability to tell when work is right, to spot AI errors and to take responsibility for decisions. In a BCG experiment, consultants using AI on a task outside its capabilities were 19 percentage points less likely to be correct than those without AI.<\/p>\n<p><strong>How do early-career professionals build judgment if AI does the entry-level work?<\/strong><br \/>\nSeek work that involves reviewing, deciding and explaining, not just producing. Early payroll data shows employment of 22 to 25 year olds in AI-exposed occupations falling behind less-exposed peers, concentrated where AI substitutes for workers, so choosing roles where AI complements your work matters.<\/p>\n<h2 id=\"sources\"><span class=\"ez-toc-section\" id=\"Sources\"><\/span>Sources<span class=\"ez-toc-section-end\"><\/span><\/h2>\n<p>Dell&rsquo;Acqua, F., Ayoubi, C., Lifshitz, H., Sadun, R., Mollick, E., Mollick, L., Han, Y., Goldman, J., Nair, H., Taub, S., and Lakhani, K. R. The Cybernetic Teammate: A Field Experiment on Generative AI and Teamwork. Organization Science, 2026, volume 37, issue 4, pages 1217 to 1242; working-paper version NBER Working Paper 33641, 2025.<\/p>\n<p>Dell&rsquo;Acqua, F., McFowland, E., Mollick, E., Lifshitz, H., Kellogg, K. C., Rajendran, S., Krayer, L., Candelon, F., and Lakhani, K. R. Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality. Organization Science, 2026, volume 37, issue 2, pages 403 to 423.<\/p>\n<p>Brynjolfsson, E., Li, D., and Raymond, L. Generative AI at Work. The Quarterly Journal of Economics, 2025, volume 140, issue 2, pages 889 to 942.<\/p>\n<p>Eloundou, T., Manning, S., Mishkin, P., and Rock, D. GPTs are GPTs: An Early Look at the Labor Market Impact Potential of Large Language Models. Working paper, arXiv 2303.10130; published in Science, 2024; occupation-level exposure data released by the authors.<\/p>\n<p>National Center for O<em>NET Development. O<\/em>NET 31.0 Database, August 2026: Essential Skills, Transferable Skills and Job Zones files.<\/p>\n<p>Deming, D. J. The Growing Importance of Social Skills in the Labor Market. The Quarterly Journal of Economics, 2017, volume 132, issue 4, pages 1593 to 1640.<\/p>\n<p>Brynjolfsson, E., Chandar, B., and Chen, R. Canaries in the Coal Mine? Six Facts about the Recent Employment Effects of Artificial Intelligence. Stanford Digital Economy Lab working paper, August 2026 version.<\/p>\n<p>Humlum, A., and Vestergaard, E. Still Waters, Rapid Currents: Early Labor Market Transformation under Generative AI. NBER Working Paper 33777, 2025, revised March 2026.<\/p>\n<p>World Economic Forum. The Future of Jobs Report 2025. January 2025.<\/p>\n<p>Hansen, M. T., and von Oetinger, B. Introducing T-shaped managers: knowledge management&rsquo;s next generation. Harvard Business Review, 2001, volume 79, issue 3, pages 106 to 116.<\/p>\n<p>Guest, D. Managers in focus as the skills gap closes: the hunt is on for the Renaissance Man of computing. The Independent, 17 September 1991, as documented in Neeley, K. A., and Steffensen, B., The T-shaped engineer as an ideal in technology entrepreneurship: its origins, history, and significance for engineering education, American Society for Engineering Education Annual Conference, 2018.<\/p>\n<p>Brown, T. Interview on T-shaped people, Chief Executive magazine, January 2010.<\/p>\n<p>Tables 1 and 2 were computed by us by joining the O*NET 31.0 skill importance ratings with the occupation-level beta exposure scores published by Eloundou et al.; the skill composites, the Job Zone 4 and 5 restriction and the median splits are our analytical choices. Tables 3 and 4 are CEOtudent editorial frameworks.<\/p>\n<hr>\n<p><em>This content was compiled with the support of AI following in-depth research, then written and prepared for publication by the CEOtudent editorial team.<\/em><\/p>\n","protected":false},"excerpt":{"rendered":"<p>The T-shaped professional, deep in one field and broad enough to collaborate, was built for a world where depth was scarce and breadth was expensive. Generative AI has changed both prices. Field experiments show AI gives individuals the cross-functional breadth that used to require a team, and our analysis of 372 degree-level US occupations finds that analytic depth without people skills is the most AI-exposed profile. This piece explains what broke, introduces a new shape, the Tree, and gives a self-audit for growing it.<\/p>\n","protected":false},"author":1,"featured_media":326042,"comment_status":"open","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[5,18],"tags":[],"class_list":["post-326041","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-is","category-strateji"],"_links":{"self":[{"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/posts\/326041","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/comments?post=326041"}],"version-history":[{"count":0,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/posts\/326041\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/media\/326042"}],"wp:attachment":[{"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/media?parent=326041"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/categories?post=326041"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ceotudent.com\/en\/wp-json\/wp\/v2\/tags?post=326041"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}