Teaching People to Prompt Is Not an AI Adoption Strategy

There is a moment many organizations will encounter as artificial intelligence becomes part of everyday work.

Leadership realizes people are already using it. Someone suggests training. Everyone agrees that training sounds responsible, and before long the conversation turns to prompts.

How do we write better ones? What words should we use? How do we get more accurate results? Can we create a prompt library for the staff?

Those are useful questions.

I teach prompting too.

But I would want to make sure we did not confuse learning how to operate an AI tool with learning how to use one responsibly inside an organization.

A person can write an excellent prompt and still make a poor organizational decision.

Prompting is a skill. AI judgment is an organizational capability.

And if we are going to prepare people to work with AI, we need both.

Good Prompting Cannot Compensate for Poor Judgment

Imagine an employee who knows how to give AI context, describe an audience, specify an outcome, request a format, and refine the response until it becomes useful.

Technically, that person may be becoming very proficient.

But suppose the information should never have entered that particular system.

Suppose the response contains a confident error the employee does not recognize.

Suppose the task involves a decision that belongs to a pastor, executive, board member, counselor, financial leader, or another person carrying organizational authority.

The prompt may have worked beautifully.

The use of AI may still have been wrong.

This is where AI training often becomes too narrow.

If our goal is simply to help people produce better output, we can teach techniques. If our goal is to help people use AI responsibly on behalf of an organization, we have to teach them how to think before, during, and after they use the tool.

What AI Can Do | What Still Belongs to Us

That principle sounds simple until we begin applying it to real situations.

If AI drafts something inaccurate and an employee sends it, who owns the communication?

If it produces language that sounds excellent but misrepresents the organization's position, who catches it?

If a volunteer creates something publicly attributed to the ministry, what review happens before it becomes public?

Now we are no longer teaching a feature.

We are developing organizational judgment.

What I Mean by AI Literacy

When I say AI literacy, I do not mean that every employee needs to understand how artificial intelligence works underneath the hood.

I mean people should understand enough to know:

what AI is useful for,

where it can fail,

what information requires care,

how to question its output,

when verification is necessary,

where human judgment remains essential,

and when to stop and ask someone before proceeding.

The evidence suggests that this is not a minor issue.

A global University of Melbourne and KPMG study found that 66 percent of respondents relied on AI output without evaluating its accuracy, while 56 percent reported making mistakes in their work because of AI. The same study found that only 47 percent of employees reported receiving AI training.

Those figures come from a global study, not specifically from ministries or nonprofits, but the leadership implication is difficult to ignore.

Access to AI does not automatically produce preparedness.

AI Can Sound Certain While Being Wrong

One reason verification matters is that AI can produce polished language that feels authoritative even when the underlying information is unreliable.

Stanford's 2026 AI Index reported that hallucination rates across 26 leading models ranged widely, from 22 percent to 94 percent on a new accuracy benchmark. The report also highlighted a troubling pattern: models performed much worse when a false statement was presented as something the user personally believed than when the same false statement was attributed to another person.

That matters because the problem is not simply whether AI sometimes makes mistakes.

The interaction itself can shape the answer.

So I would rather teach someone to ask:

"How will I verify this?"

than congratulate them because they generated something impressive in thirty seconds.

Responsible AI use requires judgment before the prompt, not merely technique inside it.

People Also Need Permission to Ask Better Questions

One mistake I would avoid is training people in a way that makes them afraid to use AI at all.

Good enablement creates clarity.

People should understand where AI is useful. They should have examples connected to their actual responsibilities. They should know what the organization permits, where additional approval is required, and what should trigger a conversation before proceeding.

They also need permission to say:

"I am not sure AI belongs in this task."

That is not necessarily resistance.

That can be good judgment.

The same is true in the opposite direction. Someone should be able to look at a repetitive administrative process and ask whether AI might reduce unnecessary work without feeling that innovation itself is disloyal to the organization.

Healthy adoption requires curiosity and boundaries at the same time.

Training Has to Reach the Actual Work

A generic AI workshop may create excitement.

That is not the same as capability.

For a ministry communications team, practical training may involve drafting, repurposing, research, review, voice, and approval.

An administrative team may need to think about correspondence, meeting preparation, information retrieval, documentation, or routine coordination.

A nonprofit development team may have different questions about grant research, source verification, donor communication, and drafting.

The technology may be similar.

The work is not.

That means training should connect AI to the responsibilities people actually carry on Monday morning.

And it should not be treated as a one time event. Tools change. Organizational expectations change. People discover new uses. New questions emerge.

AI literacy has to mature with the work.

What I Would Want Your Team to Leave With

If Chayil walked into your organization to facilitate AI training, I would not measure success by the number of prompts everyone collected before lunch.

I would want people to leave understanding what AI can help them do, where its limitations matter, what responsibilities remain theirs, what information requires caution, how to evaluate output, and what questions they should ask when uncertain.

Then we can teach prompting.

Then we can practice.

Then we can put actual work on the screen.

The objective is not to turn every employee into an AI expert.

It is to help your people become more capable and thoughtful users of AI within the responsibilities they already carry.

Before We Leave This Conversation

Once people understand how to use AI more responsibly, the next question appears naturally:

Where should we actually use it?

This is where organizations can become distracted by the technology again. Someone sees an impressive demonstration and immediately begins looking for somewhere to install the solution.

I would approach that from the other direction.

I do not want to begin with the AI.

I want to see the work.

Show me where information gets stuck. Show me what people repeat every week. Show me what depends on one person's memory. Show me where highly capable people are spending time on low judgment work.

Then we can have a much better AI conversation.

And that is where I want to pick up next.