The Lab Report — Not the MBA. Not the Bootcamp. The Other One.

Hi, I'm Charles Costa — researcher, strategist, and founder of Lexora Labs. Each week I dig into the latest research on AI, knowledge management, and the future of work so you don't have to. Here's what's on my mind this week.

Editor's note

Everyone's calling AI skills the new MBA. They're not wrong, the credential that once separated the generalist from the strategist is being repriced in real time. But the framing stops one layer too high. Treating AI fluency as a single stackable skill, like Excel or Salesforce, is exactly the wrong mental model for building durable advantage.

The skills actually commanding a premium sit upstream of the tools: knowing what to ask, how to evaluate the answer, and how to design information systems a model can reliably operate on. That's AI literacy, not AI usage, and it's been the core curriculum of information science for decades. This week's reading traces that gap, from the wage data through to the knowledge infrastructure most teams are quietly running on contaminated fuel.

This week's reading

The Fearless Future: 2025 Global AI Jobs Barometer — PwC
PwC puts a number on what founders have felt anecdotally: workers with advanced AI skills now command a 56% wage premium — double the year prior. Postings requiring AI skills grew 7.5% even as total postings fell 11.3%, and in the industries most exposed to AI, revenue per employee grew at 3x the rate of less-exposed ones.
Why it matters: The signal is unambiguous — but the same data shows the skills sought in AI-exposed roles are changing 66% faster than elsewhere, which means training to today's tool is a losing bet. You're always a product cycle behind.

If the premium is real, the next question is what employers are actually paying for.

How AI Skills and Experience Are Transforming the Workplace — WEF / Oxford Internet Institute
In the UK, AI skills now command a higher wage premium than a master's degree. Demand for AI roles grew 21% as a share of all vacancies between 2018 and 2023 — while the university-education requirement for those same roles fell 15%.
Why it matters: Employers aren't hiring credentials; they're hiring demonstrated capability to work with information at scale. That distinction is the whole ballgame, and it points straight at which skills hold value.

Knowing the premium exists is one thing. Building it on your team is another — and most upskilling programs are aimed at the wrong target.

Reimagine Learning and Development for the AI Age — McKinsey
McKinsey is blunt about the gap: most AI upskilling teaches tool usage — specific interfaces, prompts, and workflows — rather than the underlying principles that let skills transfer as the tools change.
Why it matters: Effective upskilling builds mental models, not muscle memory — teaching why retrieval works the way it does, and developing the judgment to know when to trust a model, when to check it, and when to override it entirely. That's closer to what a good MLIS program teaches than anything in a standard AI bootcamp.

And when you get the training right, the payoff shows up at a specific organizational inflection point.

Are Your People Ready for AI at Scale? — McKinsey
The companies outperforming on AI ROI aren't the ones with the biggest models or the most compute. They're the ones where mid-level employees can independently identify where AI should be applied, design the application, and evaluate whether it worked.
Why it matters: That's an AI literacy problem, not a technology problem — and it's why throwing more tools at a team without building judgment rarely moves the ROI needle.

Literacy operates on something, though — and that something is the infrastructure almost no one is auditing.

2026 Knowledge Management Predictions — APQC
APQC's read is one founders consistently underweight: the companies pulling ahead aren't treating AI as a standalone capability — they're integrating it into their knowledge management infrastructure.
Why it matters: The question isn't "do we have AI tools?" but "do we have a knowledge architecture AI can actually operate on?" Bad metadata, inconsistent taxonomy, and tribal knowledge that lives nowhere but in people's heads are the reason most KM-AI investments underperform. Before the next tool, run the knowledge audit — AI on bad infrastructure is a powerful engine running on contaminated fuel.

The through-line: the 56% premium is real, but the people who hold onto it longest won't be the ones who learned the most tools — they'll be the ones who understood information first.

If this issue sparked something — a question, a pushback, or a decision you're working through — hit reply. I read every response.

Until next Tuesday,

Charles Costa, MLIS
Founder, Lexora Labs

Reference links