Computer Tutoring Logo

What Do You Do When Someone Has Already Decided AI Isn't for Them?

You can often tell within the first few minutes of an AI training session who is enthusiastic, who is curious, and who would rather leave it well alone. A recent Copilot session got me thinking about what a trainer should do when someone has already decided AI isn't for them, and whether resistance to AI can sometimes be about something deeper than the technology itself.

Presenter entering a corporate training room

I was delivering another beginner's Copilot session, and, overall, it went well. However, there was one attendee who seemed determined not to be persuaded.

He sat with his arms folded, leaned back in his chair, and gave the impression that he had already made up his mind about Copilot and AI in general.

I completely understand that reaction. People have plenty of reasons for feeling cautious about AI. Perhaps they're worried about the technology being a threat to their own role. Perhaps they're simply tired of hearing about it. AI is everywhere at the moment, and the expectations being created don't always match the reality of using the tools day to day.

As a trainer, though, you are there to teach people how to use Copilot. You cannot really get into all the personal reasons why somebody might be reluctant to embrace the technology. You can only demonstrate what it does and let people decide for themselves.

Should you try to convert the sceptic?

That got me thinking about the role of the trainer.

If somebody walks into a Microsoft Copilot training course already convinced that AI isn't for them, should you make it your mission to change their mind?

I don't think so.

I think my job is to give people an honest picture of what the technology can do, where it is useful, where it isn't, and then let them make their own judgement.

There is also a danger in making assumptions. Arms folded does not tell me why somebody feels the way they do. It would be very easy to invent a story about somebody being frightened of AI, worried about their job or resistant to change when I actually have no idea what is going on in their life.

So I carried on with the training.

Then something else caught his attention

One of the things, I noticed, that made this person sit up and pay attention a little bit was the use of Copilot in Outlook. This surprised me as I always thought that Outlook was one of the areas where Copilot didn't really shine. Let me explain.

Imagine you've got an email conversation containing three or four messages. Rather than reading backwards and forwards through the whole thread, Copilot can summarise the conversation, pull out the actions and, importantly, give you links back to the individual emails where those points came from.

That is a fairly ordinary example of AI, but perhaps that is exactly why it works.

It isn't trying to replace anybody's expertise. It is helping somebody get through information more quickly and then allowing them to check the original messages themselves.

It's also something we cover in our Outlook with Copilot training, because email summarisation becomes much more useful when people understand not just how to generate the summary, but how to check what Copilot has actually based it on.

Now, I'm not claiming there was some dramatic conversion with this initially totally disinterested person. There wasn't.

But he did shake my hand at the end of the day.

Perhaps resistance to AI isn't always about the technology

Afterwards, I started looking into why people can react strongly to AI in the workplace.

There is some interesting research around something called AI identity threat. A study published in Electronic Markets looked at what can happen when AI starts to affect the way people see their own role and professional identity. Among the significant factors the researchers identified were changes to work and a perceived loss of status.

That doesn't mean this was what was happening in my training room. I simply don't know.

But it does suggest that resistance to AI can sometimes be about considerably more than whether somebody likes a particular piece of software.

You can read the research here: The rise of artificial intelligence - understanding the AI identity threat at the workplace.

We're also much less forgiving when technology gets something wrong

There is another interesting piece of research that has stayed with me.

Researchers studying what they called algorithm aversion found that people could lose confidence in an algorithm more quickly after seeing it make a mistake, even when the algorithm performed better overall than a human forecaster.

The research wasn't about Microsoft Copilot or today's generative AI tools, but the idea still made me think.

If a colleague gets something wrong, we normally accept that people make mistakes.

If AI confidently gives us one ridiculous answer, our reaction can very quickly become, "See, I knew this thing was useless."

You can read that study here: Algorithm Aversion: People Erroneously Avoid Algorithms After Seeing Them Err.

Maybe expertise isn't disappearing. Maybe it's moving.

The Outlook example also made me think about expertise.

There is a fear that AI means people won't need to know as much because the machine will simply give them the answer.

I'm not convinced that is what is happening.

If anything, the ability to judge the answer may become more important.

You still need somebody who understands the subject well enough to say, "Yes, that makes sense", "No, that doesn't", or "Where exactly did that information come from?"

Perhaps the role of the expert isn't disappearing. Perhaps it is changing.

Maybe the experts of the future won't always be the people who can remember the most information. They'll be the people who know what to trust, when to question it and how to apply it.

So what would you do?

I came away from the session thinking that successful AI training doesn't necessarily mean everybody leaves the room converted.

Sometimes it may simply mean that people have seen enough to make a more informed decision for themselves.

And perhaps that's the best a trainer can do.

If your organisation is still unsure about where Copilot genuinely fits, our 90-minute Copilot training sessions are designed around practical tasks in apps such as Outlook, Excel, Word and PowerPoint rather than trying to convince people that every problem needs AI.

I would like to finish this article with a question? If you, as an AI champion or user, had to convince someone of the benefits of AI. What would you do?

Thank you for subscribing. You'll receive our newsletter soon.