New research confirms what frontline workers already knew: AI deployed without human orchestration erodes trust faster than it delivers efficiency.
A recent study from SHL, reported by Kit Eaton in Inc., reveals a stark divide between leadership enthusiasm for AI and workforce reality. Only 27% of workers fully trust their employer to use AI responsibly. Nearly three-quarters say being interviewed by an AI agent would harm their perception of the company. Six in ten don’t want AI evaluating their performance or making career-impacting decisions.
These aren’t Luddites. Nearly half of these same workers actively want to upskill in AI. They’re not rejecting the technology—they’re rejecting how it’s being deployed.
Workers don’t fear AI. They fear being judged by a machine that misreads a pause as doubt.
The Missing Layer
What the data reveals isn’t an AI problem. It’s an orchestration problem.
Organisations are treating AI as a tool to deploy rather than a collaborator to orchestrate. They’re automating decisions without building the human-in-the-loop architecture that earns trust. Workers feel it immediately—the interview that moves too fast, the performance review that quotes metrics without context, the promotion algorithm that can’t see what happened off-camera.
As the SHL research notes, people appreciate AI’s speed but remain unconvinced it has “good judgment without human involvement.”
That phrase—“without human involvement”—is the key. It’s not AI that workers distrust. It’s unmediated AI. AI making decisions in a black box.
Why the Existing Playbook Fails
Most organisations respond to AI distrust with communication. Town halls. FAQs. Reassuring statements about “responsible AI” and “human oversight.”
It doesn’t work.
Workers aren’t asking for better messaging. They’re asking for visible methodology. They want to see how AI fits into decisions, not just hear that it does. Policy statements don’t build trust. Architecture does.
AI without oversight isn’t innovation. It’s abdication.
This is precisely the gap COMINDING exists to close.
Orchestration, Not Automation
COMINDING treats human-AI collaboration as a deliberate practice requiring structure, transparency, and—crucially—friction.
The framework uses multiple AI platforms strategically: different tools for creation, critique, data synthesis, and verification. But the human remains the orchestrator, not a passive recipient of AI output. Every AI contribution passes through human judgment before it shapes decisions.
This isn’t inefficiency. It’s architecture.
When you use COMINDING, you can articulate exactly which platform contributed what insight, where perspectives conflicted, and how human judgment resolved those conflicts. That’s the transparency the 73% of distrustful workers are demanding. They want to know where and how AI is used—and COMINDING provides that visibility by design.
Disagreement as Feature, Not Bug
One of COMINDING’s core principles is deliberately seeking AI disagreement. When Claude contradicts ChatGPT, when Gemini surfaces data that challenges Perplexity’s citations, that friction isn’t a problem to solve. It’s a signal to investigate.
In practice, this looks like catching a flawed assumption before it escapes into production—because two AI platforms disagreed and the human had to arbitrate.
Contrast this with organisations deploying AI as an authoritative decision-maker. No friction. No competing perspectives. No human checkpoint before the AI’s output becomes policy.
Workers intuitively understand the danger here. They know that AI trained on historical data can perpetuate biases. They know that algorithmic efficiency isn’t the same as good judgment. They know—because they live it daily—that context matters in ways AI systems struggle to capture.
COMINDING codifies what workers already sense: AI output needs to be challenged, not accepted wholesale.
Trust Is Infrastructure
The SHL research carries a warning for leaders racing to deploy AI: “Bang the AI drum too loudly, without showing that you plan to use it responsibly and with plenty of human-in-the-loop supervision, and you risk shooing away potentially valuable job candidates.”
Trust isn’t a soft metric. It’s infrastructure. Erode it, and your transformation programmes fail—not because the technology doesn’t work, but because the people who use it have checked out.
Policy statements won’t fix this. Visible methodology will—showing workers how AI fits into decisions, with deliberate orchestration they can see and understand.
The Path Forward
The workers in this study aren’t asking for AI to disappear. Only 21% want that. The rest are asking for something more nuanced: AI that enhances rather than replaces human judgment. AI with clear boundaries. AI deployed transparently, with humans visibly in the loop.
That’s not a rejection of AI. It’s a demand for AI done properly.
COMINDING answers that demand. Not by slowing AI adoption, but by providing the collaborative architecture that makes adoption sustainable. When workers can see how AI contributes to decisions—and where human judgment intervenes—trust becomes possible.
The alternative is what we’re seeing now: a workforce increasingly suspicious of tools their leaders are increasingly enthusiastic about. That gap doesn’t close on its own. It requires intentional design.
Orchestration isn’t optional anymore. It’s the only way trust survives the next wave of AI.
COMINDING is a framework for human-AI collaboration that treats multiple AI platforms as collaborative partners rather than authoritative oracles. Learn more about the methodology and its principles throughout this site.