Don’t bet on AI to level up your B players
AI is already improving productivity. Software teams are shipping faster, customer-support teams are resolving more issues and sales professionals are producing tailored proposals in hours rather than days.
Finance teams are underwriting opportunities with greater speed and, in many cases, greater accuracy.
The gains are real, because AI can rapidly connect, synthesize and apply information across data sources, systems and prior work that once required substantial manual effort. But it's not a talent equalizer.
To the skilled, AI is a capability amplifier – accelerating research, analysis, decisions and execution.
In unskilled hands, it can become the junk-food version of professional judgment: prepackaged, overly processed, superficially satisfying and lacking the nutritional value of original thinking, discernment and real expertise.
To the skilled, AI is a capability amplifier – accelerating research, analysis, decisions and execution.
"Software engineering gave us the first glimpse of what's now unfolding across every profession. The best engineers spend less time writing code and more time orchestrating AI coding agents," says Jill Tarallo, COO and AI Transformation Officer of Rollick, a custom CRM provider for recreational equipment manufacturers and dealers.
"They define the work more clearly, provide richer context, anticipate trade-offs, challenge assumptions and continually refine the solution toward a better outcome. Give two developers the same AI coding agent, and the results are rarely comparable.
"A Players recognize the difference between code that works and a solution that's right for the business. B Players are more likely to accept the first plausible answer."
Starting with the A Players
The Scaling Up Performance Platform begins with the right people because AI is an accelerator of capability – not a replacement for A Player judgment, accountability and execution discipline.
That distinction has profound organizational implications. When A Players can produce materially more high-quality work with AI, it compresses production. Work that once required multiple handoffs, review cycles, meetings, administrative effort and layers of routine execution can now be completed by fewer people in less time.
Layers flatten and roles consolidate. AI does not simply help people do more; it raises the standard for high performance while reducing the number of people required to deliver it.
As that production compression accelerates, gaps in talent quality become impossible to hide. Leaders begin to see the difference between people who create meaningful value – solving hard problems, exercising sound judgment, improving decisions and moving critical priorities forward – and those who primarily generate activity, process and output without materially advancing the business.
That is why Scaling Up founder and author Verne Harnish and I are advising CEOs that the winning move is to build superteams of A Players using AI.
The objective is to pay fewer people better: use AI to expand the capacity of exceptional talent, then reinvest the gains to attract and retain even more A Players. Done well, that turns what could become a vicious cycle of talent displacement into a virtuous cycle of higher performance, stronger economics and greater enterprise value.
The widening gap between A and B Players
John Kim, Co-Founder and CEO of Paraform, recently challenged the standard narrative that AI’s primary impact on jobs will be the hollowing out of entry-level work. His warning is more consequential for CEOs: AI is making "good enough" workers redundant while concentrating demand around a much smaller group of exceptional people.
The market is not merely eliminating junior work; it is raising the premium on A Player-level judgment, initiative and output. The gap between those groups is widening faster than most leaders realize.
AI may make a B Player faster. It may help them generate more output, assemble more information and package work more convincingly.
When A Players can produce materially more high-quality work with AI, it compresses production.
But it will not reliably give them the judgment, initiative, pattern recognition or accountability of an A Player. More often, it makes the gap impossible to miss.
That is the dangerous illusion many leaders are beginning to believe – that AI will raise the floor and level up B Players.
It may raise activity, improve presentation and increase volume. But it will not turn average judgment into exceptional performance.
The evidence is already clear. In ‘Navigating the Jagged Technological Frontier', researchers studied 758 Boston Consulting Group consultants – an elite talent pool, not a stand-in for true B Players.
The lower-skilled consultants who benefited most from AI were better described as A Potentials: capable people still developing the judgment, experience and pattern recognition of their top-performing peers.
On structured tasks within AI’s reliable capability range, GPT-4 users completed 12.2 percent more work, did it 25.1 percent faster and produced higher-quality results. But on a complex strategic task requiring subtle synthesis and independent judgment, AI users were 19 percentage points less likely to reach the correct answer.
That is the real dividing line. AI can be extraordinary inside a structured, information-rich problem.
Outside that range, people need enough expertise to recognize when the answer is wrong, incomplete or based on a faulty premise.
The cost of convenience
Borrowing a phrase from the technology documentary The Cost of Convenience, the AI version is equally clear: convenience is rarely free.
The cost is not merely more AI slop – synthetic, empty-calorie, faux-knowledge and frankly thoughtless work that sounds informed enough to circulate, yet lacks the soul, deep thinking and discernment required to make the right call. Over time, convenience can also weaken the independent judgment required to catch what AI misses.
Medicine offers a useful warning. In the Stanford-led randomized clinical vignette study, ‘Influence of a Large Language Model on Diagnostic Reasoning', ChatGPT alone achieved a median diagnostic-reasoning score of about 92.
Physicians using conventional resources scored 74, while physicians with ChatGPT access scored 76 – no meaningful improvement.
The market is not merely eliminating junior work; it is raising the premium on A Player-level judgment, initiative and output.
The takeaway was not anti-AI. The model was highly capable, but simply handing a professional a powerful tool did not automatically create a better human-plus-AI result; physicians needed better training and integration to use it well.
A second warning, documented in the article, ‘New Study Suggests Using AI Made Doctors Less Skilled at Spotting Cancer', came from The Lancet Gastroenterology & Hepatology.
After routine AI exposure, doctors’ unassisted adenoma-detection rate fell from 28.4 percent to 22.4 percent – a six-percentage-point absolute and roughly 20 percent relative decline.
The study was observational, so it does not prove AI caused the decline. It does, however, raise a serious question about deskilling: AI can help with the work in front of you, but dependence on it can quietly weaken the capability behind you.
The counterpoint is equally important. A 2026 Science study led by Harvard Medical School researchers found that OpenAI’s o1 model performed exceptionally well on text-based physician-reasoning tasks.
In one test involving 76 emergency-department cases, it produced correct or near-correct diagnoses 67 percent of the time versus roughly 50–55 percent for attending physicians using the same written records; with fuller information, it reached 82 percent.
The study is a powerful argument for AI in expert hands and within structured, text-rich work. It was not, however, a live bedside trial – and it did not prove that merely adding AI to a clinician automatically improves the human-AI operating model.
AI plus expertise is the key
That is the business lesson. The advantage is not AI alone, but an A Player who knows how to model data, test it, challenge it, contextualize it and train it.
"While AI has massively delivered on its production promise, the real opportunity now is scaling human judgment in AI usage," says Marty King, CEO of AIX, an AI implementation consultancy.
We see these judgment faux pas with B Players all the time. Within the Scaling Up Performance Platform framework of People, Strategy, Execution and Cash, we see the most business malpractice around AI in people systems.
B Player managers are abdicating their responsibilities, writing AI-generated performance reviews filled with thoughtless, saccharine AI slop and passing them off as coaching.
Convenience is rarely free.
Strategy is the next worst offender, with voluminous, barely readable documents passed off as strategy despite offering little human insight.
Execution is a little better, but watch for AI-generated quarterly priorities or rocks that managers cannot articulate or deliver on when pressed.
Cash is actually the best, with custom agentic AI models and graphical sensitivity tools drawing on data lakes to deliver complex financial modeling and analysis at the speed of thought. This is because financial data is structured and easily mapped.
Unless it is harnessed by an expert user with disciplined models, rules and operating standards, AI is notoriously inconsistent in both output and format.
That gives the AI Tourist just enough confidence to become a Template Cowboy – peppering the organization with differently formatted, poorly articulated people and strategy documents.
Now you're wasting tremendous time navigating new formats, terms and conclusions every time. Meanwhile, the B Player silently congratulates himself for being clever, and you're hesitant to rein him in because everyone is telling you to encourage AI adoption in your company.
Driving consistent gains through discipline
One of the fundamental tenets of Lean principles is consistency. Undisciplined AI use violates that principle by creating random variation at scale – new formats, new terms, new logic and new confusion every time someone gets clever with a prompt.
This is not innovation. It is overtly counterfeit work and production mutilation on steroids.
"Give an A and B Player the same AI tools, and you'll learn far more about the people than you will about the technology. That's why organizations with exceptional talent continue to pull away from the pack," Tarallo says.
"It's literally changing the economics of talent. The B and C Players are actually outing themselves with AI!
Unless it is harnessed by an expert user with disciplined models, rules and operating standards, AI is notoriously inconsistent in both output and format.
"You just gave that mediocre player the superpower to amplify their mediocrity. You think they must be smart, but you just aren't sure why they don't get it.
"Now with AI, just turn them loose with a GPT and you'll know for sure why they don't. It's like a new x-ray vision to immediately understand what is wrong with them – whether it is a lack of skill, will or smarts!"
Heritage Hill Capital Partners, a Cincinnati-based multifamily real estate investment firm and operator, has built sophisticated AI models and agents into its operating discipline. It uses them to make smarter, faster decisions on real estate selection and operations to bolster investment returns.
"AI is a multiplier, but multiplication has an unforgiving property: anything times zero is still zero. That's the part leaders miss when they hope AI will carry a B performer," Heritage Hill CEO Sean Blythe says.
"If the judgment isn't there, no amount of processing power can create it. You simply get a facsimile of competence, delivered on time and in a fancy format."
At Scaling Up, we coach companies to run with the authenticity and precision of a Swiss watch. That is why our early agentic AI development started with people – because most companies still have too much business malpractice around talent identification and development.
Our Scaling Up A Players agent helps managers produce higher-quality job scorecards, performance profiles and A Player agreements by forcing better thinking before AI does the synthesis.
Think punch-card programming: the expert had to be clear and disciplined with the inputs, and the computer handled the processing.
These powerful, consistent and – candidly – gorgeous documents are fundamental to coaching A Players, who appreciate the clarity and expectations they establish.
When managers are challenged with structured prompts around performance standards, outcomes, lead measures and role expectations, searches improve, A Players are identified faster, onboarding becomes quicker and more effective, and KPI attainment improves, even among incumbent employees.
For CEOs, the move with AI is straightforward: do not use it as an excuse to tolerate B Player thinking on steroids. Use it to raise your work standards.
- Apply Lean principles: reduce variation, standardize what matters, build quality into the process and eliminate wasteful rework. Build clear Job Scorecards tied to outcomes.
- Insist that managers understand the work they coach, the lead measures that drive results and the causal connection between activity and performance.
- Create quality gates around decisions, customer communication, hiring, forecasting and strategy.
- Demand better than lazy, meandering, barely coherent AI-generated strategy documents produced faster than brewing a cup of coffee.
Scaling Up gives CEOs the operating system to do exactly that across People, Strategy, Execution and Cash. AI can accelerate a disciplined system, but it cannot replace one.
Do not bet on AI to level up your B Players. Instead, build a superteam of A Players using AI – and let the technology multiply what is already excellent. Scaling Up has all the tools and expertise to make this happen in your organization.
Want to build a high-performing team and unlock AI's full potential? Click here to connect with a certified Scaling Up coach to help accelerate your business growth.