The new divide at work: judgment, not just adoption.
The big split in companies is no longer remote versus in-person, or managers versus staff. It is between people who run AI with good judgment and people who either avoid it or use it without checking. This brief lays out what the research says, including the parts that do not flatter AI.
The baseline is moving
Read those together. Almost every company is investing. Almost none have it working well. That gap is where most of the risk, and most of the opportunity, sits.
What speed buys you
In a study of 5,179 customer-support agents, an AI assistant raised the number of issues resolved per hour by about 14%, and by about 34% for the least experienced workers (Brynjolfsson, Li and Raymond). That is a real gain, in a structured job with fast feedback.
What speed does not fix
In a study of 758 consultants, people using AI did better on tasks inside the tool’s strengths. On a task outside them, AI users were about 19 percentage points less likely to get a correct answer than people working without it (Dell’Acqua and colleagues). The AI sounded sure. It was wrong. Nobody checked.
There is hard news too. A Stanford study found about a 13% relative drop in employment for workers aged 22 to 25 in the jobs most exposed to AI. And a 2025 World Economic Forum survey found 41% of employers plan to reduce staff where AI can do the work.
The honest summary: AI helps most where work is structured and checked. It hurts where people trust it blindly. It does not repair a broken process, a vague strategy or weak management.
Four judgment calls that matter
Real AI skill is not clever prompts. It is knowing:
- When to automate a task that repeats the same way every time.
- When to verify the machine’s answer before anyone relies on it.
- When to keep a person in the loop for a decision that affects someone.
- When your data must never go in. Customer records, passwords, health details and anything your company calls private.
What the better-run companies do
- Train people on purpose. They do not expect staff to figure it out alone.
- Pick specific workflows. They do not hand out tools and hope.
- Set guardrails first. They decide what data is off limits before an incident forces the question.
Without those three, “use AI” turns into “work faster and cheaper.” You get fast, unchecked output and quiet, unmanaged use of AI tools that nobody approved.
Where C.G.I. fits
Human judgment is the advantage. The G.A.I.N.S. certification teaches three skills any employee can use on Monday: ask clearly, check the answer, protect the secrets. The weekly AI Win turns that into a habit your company can measure. The Company Brain keeps what you learn.
We do not promise raises, promotions or savings. We promise a clear standard and a way to see whether your people are meeting it.
Sources
- McKinsey & Company, Superagency in the workplace (2025): 1% of leaders call their companies mature in AI deployment.
- World Economic Forum, The Future of Jobs Report 2025: 39% of existing skills transformed or outdated 2025 to 2030; net +78 million jobs by 2030; 41% of employers plan workforce reductions where AI can automate tasks.
- Brynjolfsson, Li and Raymond, Generative AI at Work (NBER): 5,179 customer-support agents; about 14% more issues resolved per hour, about 34% for novices.
- Dell’Acqua et al., Navigating the Jagged Technological Frontier (Harvard Business School working paper, 2023): 758 consultants; outside the tool’s frontier, AI users were 19 percentage points less likely to be correct.
- Stanford Digital Economy Lab (2025), Canaries in the Coal Mine: about 13% relative employment decline for ages 22 to 25 in the most AI-exposed jobs.
Most of these are observational or single-setting studies. They show patterns, not guarantees for any one company.