#588 The AI-Human Issue: Why Giving AI a Face Backfires | Niels Brabandt
The AI-Human Issue: Why Giving AI a Face Backfires
By Niels Brabandt
"Here is your new AI process. His name is Karl." Are you happy about that? Most people are not, and as a leader you should ask yourself why. I am Niels Brabandt, and in my work as a consultant, trainer and interim manager I see the same pattern again and again: a technology that is useful, helpful and coming whether we like it or not meets massive resistance inside organisations. That is the AI-human issue, and it is getting worse.
The Paradox: Helpful at Home, Threatening at Work
Most people already use AI in their private lives. They ask it for a restaurant recommendation, a travel plan or a quick answer, and they say it is genuinely helpful. The same people walk into the office on Monday and say: "At work I am not so sure. Maybe it takes my job."
The technology is not the problem. People like what AI does. What they fear is what their organisation might do with it. That distinction matters, because it tells you where the resistance really comes from: not from the tool, but from the way leaders introduce it.
The Business Case Is Rarely the Problem
Ask anyone whether they enjoy expense reports or the paperwork around a single day off. Nobody does. I regularly hear people say that the time they spend in the software requesting a day off costs more than the day off itself.
Now do the arithmetic. Take an organisation with 400 people, 150 of them in the field: technicians, maintenance staff, sales. If only 50 of them spend 10 hours a month on expense reports, that is 500 paid working hours every month, spent moving money from A to B and back to the people who spent it in the first place. Not a single client contacted, not a single product sold, not a single innovation brought forward.
Offer people automation for that, and most of them will say yes immediately, because it frees them for the work they are actually paid for. AI goes further than classic automation. The idea is welcome. The execution is where it fails.
When the Wrong Ideas Get Airtime
In one interim mandate, we introduced AI properly: a general introduction, a few training sessions, then an open call for ideas. Many suggestions were excellent. One was not. Someone proposed taking regular screenshots of employees' screens so AI could track who works on what, how quickly and how efficiently.
First, in most countries this is illegal, for good reason. Second, it would be psychologically harmful. Third, trust would hit a record low and nobody would speak openly again. And people would quickly use AI to pretend to work. More control rarely produces better work.
The most revealing moment came one level higher. A senior leader liked the idea and suggested extending it to middle management. Then we discussed what that would mean: tracking how productive their meetings were, whether promises made in annual appraisals were kept, how often time was wasted. Suddenly the proposal was a privacy problem. Privacy for some, but not for others. Your people notice that double standard immediately.
The Real Mistake: Humanising AI
Here is a real case from a mid-sized company with a few hundred employees. The leadership team had the best intentions. At an all-hands meeting they presented their new AI processes. Expense reports filed with AI, money in your account the next day: applause. Time off of less than two weeks approved automatically within 24 hours: applause. Client voice notes transcribed and categorised straight into the CRM: massive applause.
Then they presented the processes themselves, each with a face and a name. The HR process was "Paula", marketing was "Francine", finance was "Karl" and IT was "Tim", all illustrated with AI-generated portraits.
Within minutes, hands went up. Why do HR and marketing carry female names while finance and IT carry male ones? Why are all the names English? Why are all the faces white? The discussion that followed had nothing to do with productivity. It was entirely avoidable.
And there is a deeper contradiction. You tell people: "Don't worry, AI is not taking your job." In the same breath you show them human faces and names, people who look like them and who now perform parts of their work. Then you wonder why there is resistance. Any qualified consultant or organisational psychologist could have predicted this outcome.
What Works Instead
One of my clients, whose details I have changed here, runs a core process called ACAT. The first A used to stand for "automated". Today it stands for "AI-assisted". No face, no persona, no gender, no cultural signal. People simply say: "That is the process." Adoption was smooth, and people love it.
The psychology behind this is well established. Sense-making is a basic human need. When people cannot see why something makes sense for them, they will create their own explanation, and that explanation usually takes a negative turn. It then spreads through the organisation. I call it toxic spread.
Involvement is the antidote. Announce that management and IT are holding a retreat and "AI is coming", and you trigger panic.
Instead, sit down with representatives from every department and ask: Which tasks are monotonous? Which tasks should you not be doing at all? Where are you too expensive for the work? Those representatives will go back to their teams with good news: the accounts bookings, the fight over meeting rooms, the 40-page documents for the authorities are now drafted automatically. Nobody writes 40 pages any more. People only review them.
Legitimacy Is at Stake
When leaders humanise AI, employees conclude that management has no idea what they think, feel or worry about. That is how a leader loses social legitimacy. Once you are socially delegitimised, people may still pretend to follow you, but they will not.
If you want support for AI, design it with your people, name it neutrally and position it for what it is: assisted intelligence that removes the burden so people can do excellent work again.
What Leaders Should Do Now
1. Start with the burden, not the tool: identify monotonous, repetitive tasks that nobody is paid to do and quantify the hours they cost.
2. Involve representatives from every department in the design before anything is announced.
3. Give AI processes neutral, functional names. No faces, no personas, no human names.
4. Rule out surveillance use cases explicitly and apply the same standard to every level of the hierarchy.
5. Explain the purpose early and clearly, so people do not have to fill the gap with their own worst-case story.
If you would like to discuss how this applies to your organisation, I am happy to hear from you.
Apply, apply, apply.
Niels Brabandt is an expert in Sustainable Leadership with more than 20 years of experience in practice and research.
Niels Brabandt
---
More on this topic in this week's videocast and podcast with Niels Brabandt: Videocast / Apple Podcasts / Spotify
For the videocast's and podcast's transcript, read below this article.
Want your leaders to deliver better results, sustainably?
Excellent leadership makes the difference.
Let's talk: NB@NB-Networks.com
Answer guaranteed within 24 hours.
Contact: Niels Brabandt on LinkedIn - https://www.linkedin.com/in/nielsbrabandt
Website: www.NB-Networks.biz
Podcast and Videocast Transcript
Here is your new IT/AI processor, and it's named Karl. Aren't you happy about that? And quite often people are not. And the question is: why is this the case? Why does this happen? A technology which is there, which is useful, which is helpful, which will come anyway if you like it or not, often is met by massive resistance.
And the question is: where does this come from? What is the AI-human issue? And we are going to talk about exactly that, because this is an ongoing aspect of what we see right now in AI technology. I'll give you a very simple example. Let's say you have a certain thing to do. And let's face it, most people at home use AI, whether it's for a restaurant recommendation, or for a travel plan, or for anything you like to do quickly.
Many people say, "Hey, AI is really helpful," while at the same time the same people say, "Yeah, well, at work I'm not so sure, because maybe, maybe they're taking my job. Isn't that possible? Isn't that entirely possible that they take my job?" Well, we now have to talk about where are we, and how to handle it. The AI-human issue is strong, and it's really getting problematic.
So what's the situation here? Let's say you have a certain process. When I tell you, "Hey, do you know expense reports?" most people say, "Oh please, stop. Oh God, no." Or maybe remember when you want to have a day off, the paperwork around that. You like it? People really say, "I don't get paid for this." I'm sitting there in this software for 30 minutes trying to figure out how to actually get a day off. This is more expensive than the actual day off just trying to get it. How is this? And here we go.
So we see many issues where technology suddenly is helpful. If anyone would say, "Hey, want to have expense reports automated?" most people would say, "Yeah." And with AI, it's even better because it's more than automation. And when I say, "Hey, do you want these kind of processes where you only sit around, do the same thing monotonously hours and hours?" most people say, "Yeah, because then I can finally get to the work I get actually paid for." So the technology is helpful. The main problem here is that often the implementation of the technology and the way it is done is not.
The question is: how can this happen? Let's say there's a certain idea. Let's say you have an idea and say, "Hmm, we need to get rid of expense reports because we have 400 people working here. About 150 are in-field, either people who are in the technology sector, who are in the maintenance sector, or in the sales sector." 150 people basically sit around in front of expense reports 2 hours per week, or maybe 10 hours a month. And let's say even if it's only 50 people, that's 500 working hours paid for simply filing something for an internal administrative task to shift money from A to B within the organisation back to the people who spent it in the first place. Not a single client contacted, not a single product sold, not a single innovation brought forward.
So the idea, the idea of saying we actually get a different solution in place, it's very much welcome. The problem now is that often the execution is very poor. And often the execution fails because some people are heavily involved and others are not. I'll give you an example. One situation which I had in an interim mandate. We said, "Hey, this is what AI can do: general intro, bit of training, couple of training sessions, and then we say, 'Please give your input forward.'" And some ideas were amazingly useful.
But some ideas, you really wonder, where do they come from? One of these ideas was that someone says, "Well, I saw that AI can look on screens." So what we should do is, we should take screenshots from people's screens, because then we can see who is working on what, what are they do, who is working, what are they doing, who is working quickly, who not so much, and why, who's actually very efficiently working, and who's probably not working that much. We can track all of that.
So first things first. In most countries, this is illegal, for a good reason. Second, it would be massively psychologically harmful to people. Third, trust would go on a record low. No one would say anything anymore. And by the way, people will then figure out how to use an AI to pretend that they are working. So this is not how anything. We're putting more controls in, in almost any case. Not all, but almost any case. Does not lead to higher quality of work. In some cases it does, in many it does not. So the concept of more surveillance.
And by the way, the funny thing was that one level higher from the leadership said, "Oh yeah, I think this is pretty good. I think we should do that with middle management as well." How often did managers actually sit in meetings? Were these meetings productive? What did they contribute? Was their contribution well proposed, well presented, well prepared, or were they just winging it saying, "Oh, we need to be strong on the matter and talk to our customers because our customers pay our salaries"?
And by the way, how often, actually, during the annual appraisal interview, did leaders either allude to something or promise something and then didn't keep the promise, or making up cheap solutions? How often, in any meeting, was time wasted? How often did they talk trash about higher-level management? We could track all of that. And suddenly the leader proposing that concept wasn't too fond of it anymore. What a surprise. And by the way, he then said, "Oh sorry, that's privacy, isn't it?" Okay, so privacy for you, but not for others. What an interesting concept, to say the least.
The point here is, IT can be tremendously helpful. As soon as I take things off your desk that genuinely are a nuisance for you. General admin, for example. Most people do not have a contract where it says, "You get paid for general admin," because that's what makes the company go around. No, that's something that needs to be done. But especially monotonous, repetitive tasks can be done very well with AI.
The problem now is, the problem now here is, and now I'm going to tell you something, this is a real-world example, and I say something upfront. It's a mid-size company, a couple of hundred people working there, and they presented an AI project, and they had the best things on their mind. These are genuinely good people. It's a mid-size business. The way how they executed it, however, didn't go down too well. I'll give you an example why.
So there was a general meeting, hundreds of people in the room saying, "Hey, we now have amazing AI processes in place." And the first thing they said, "Remember these expense reports? Think of the past. You can file them with AI, you get the money next day." Round of applause, massively positive.
Then someone says, "Remember you wanted to have a day off, and then you had to actually propose a day off, and then we had a workflow, this, this, this, this." Anything of less than 2 weeks' time off gets confirmed automatically within 24 hours or less, so you don't have to wait for weeks. Massive round of applause, massively positive.
Then they said, "This is voice notes for clients. This goes into our CRM system. You don't have to write things anymore. You can speak them in," and they get categorised in different items. All of it automatic. Massive, massive round of applause.
So people think, "This is really useful," and then they say, "And here we present our processes." And then they show it. And I'm not making this up now.
Then they said, "Okay, this is our HR process." Suddenly you see a face on the slide. For whatever reason, you give faces to AI tech processes. I have no idea why people do this, but many do it because I think they have something good in mind that people will like it. Most people don't, and I'll tell you in a minute why.
So they said, "Oh, we have our HR process that is Paula." And they were like, "Uh, okay." Looked at the picture. Slightly disturbing AI-generated picture. Okay.
"Oh, and we have our marketing process. Our marketing process is Francine." And they went, "Okay, another face, another okay."
"Oh, by the way, and we have our finance process. This is Karl."
And before someone could say anything, they'd, "Oh, we have our IT process. Our IT process is Tim." And then people raise their hand and say, "May I ask, why are certain departments like HR and marketing with potentially more female-identifying names than people assigned?" And then you have finance and IT with more male-representing people and the names assigned. And by the way, why are these all English names and no names from any other country? Why do we only see white people? And why do they look? Massive discussion around this.
And by the way, so this is called humanisation of AI. You want to convince people that AI is not taking away their job. So you say, "No worries, AI is not taking your job." And by the way, look at these faces, people, and names that totally look like you and are totally humanised. And by the way, these are now taking part of your jobs. And then you wonder why there is resistance towards the implementation of AI. Who could have thought? Except, probably, anyone if you had asked maybe professional consultants, psychologists, or anyone who basically thinks, "I have a tiny bit of qualification in that space of leadership," and then they could have told you immediately.
And by the way, there are great implementations. One of my company, one of my clients, and I'm now changing the name, they had a process called ACAT. So A was for automated, and then there was a very, very industry-specific, company-specific term. And it's still ACAT, but it's not auto, it's AI-assisted. The A shifted from automated, automatic, to AI-assisted. And people loved it because ACAT isn't a name, there was no face to it. People simply went, "This is that process." Name it as you like, but when you humanise AI, you are asking for a discussion that you do not want to have.
From names to faces to lookalikes to cultural diversity to other diversity aspects to people being offended, by the way, rightfully so. For people making question marks about why do you do this, what is your plan behind this, why do you humanise all of that, and what is our role here in the future. All of that starts immediately. When you say, "Hey, we take the general admin off the table, and here are a couple of abbreviations you need to remember when you activate that process." Perfect. You get a round of applause. Anyone moves in happily ever after.
So very important is the implementation of all of this. You have to step into reality that if you do not prevent people from having that thought, the overthinking process is real, and sense-making is a psychologically determined fact. When people do not see that this makes sense to them, they will make it make sense. And they usually take a negative turn on it, and they take the negative turn, spreading it all over the organisation, called toxic spread. So this is the reality we have here.
From there, as anything, you need to include people in the design. I'll give you a very simple example. When you say, "Hey, you know, management and IT are having a retreat, and yeah, in a couple of months you're going to hear about this IT thing. You know, AI is coming," and buckle up. Massive panic from there, massive resistance. When you say, "Hey, we sit down with representatives from each department," and they say, "What are the tasks that are monotonous? What are the tasks that you think you shouldn't be doing? What are the tasks where you actually are too expensive for us to do them? What are the tasks you think shouldn't even be your tasks that we somehow can put into AI?"
And many people, very quickly, will find something where you say, "Hey, I think this is it." And very quickly, these people will then go to the departments and say, "People, people, people, remember the whole accounts thing which we had to book? Gone. Automated. Remember the fight around meeting rooms? Gone. It's all automated. Remember all these searching through documents with clients, and then we have to put it into a certain thing to file it with a certain authority? Done. The whole document is automated. We only have to look through it after it was written."
So we don't have to write 40 pages. We only have to read 40 pages, and basically only in a certain couple of departments because most of that data is automated, correct data anyway. People will like it when you implement something with a design process of which they are part. But, and I have to say but here, prevent to humanise AI. Stay away from humanising AI. It might look nice to you, and it might feel like a ring to you, and that usually is.
And by the way, that also sends a very clear sign. People see very clearly when they say, "He doesn't even see our perspective. He doesn't even know what we're feeling. This manager, he, she, they have no idea what we are thinking, feeling, what our sorrows and worries are. Our life situation seems to be completely unknown to them." And then again, you lose your social legitimisation. Your social legitimisation, you get socially delegitimised. And that's the worst thing for a leader to happen. As soon as this happens, you're done. That was it. No one's following you. Even if they pretend that you follow, that they follow you, they won't.
So very important is, stay away from naming your AI like humans, because when you think that this is a positive thing, most people will say, "Now it really looks like it's taking our jobs, and that doesn't feel great for me." Be aware of that. Do it as we discussed, because when you implement it in the way that we just discussed here, everything will be better from there. And I wish you all the best implementing this in your organisation.
And when you now say, "Oh wait, I have 1, 2, 3, 18 questions about this," feel free to contact me anytime. So first, of course, when you just watch me on YouTube, feel free to leave a like there. Of course, you can also subscribe to my channel, leave a comment there, always happy to discuss. And of course, when you listen to me on Apple Podcasts or Spotify, always happy to receive a recommendation from you or a review, five stars. Thank you very much for doing so.
By the way, this channel only grows by recommendation. We are now in our sixth season, sixth year with more than 500 episodes in English and German available. And we grew to international size. Thank you very much for doing so. This would never have happened without you. We only grow by recommendation. There is zero, absolutely zero advertising or paid. There is zero paid outreach. It's only by recommendation and quality content. Thank you very much for your year-long support.
When you now say, "Do you have more than this weekly episode?" Yes, we do. The YouTube Shorts, as the name says, short leadership tips, either from me or from people with external proven expertise, hand-selected experts. These YouTube Shorts can always be seen, as the name says, on YouTube. So it pays off not only to subscribe to the channel, but put the tiny bell in there, then you do not miss out on anything which we release. Usually, you get one tip per day. I promised you one tip per day. So far, we are doing well, and we are well ahead of plan.
And of course, you can also follow me on Apple Podcasts, Spotify, or when you say, "Look, we need something really specific. We need a trainer, a speaker for our event, for our convention, a coach, a mentor, a consultant, a project or interim manager." That's what I do for a living. Feel free to visit my website, nb-networks.biz. Of course, when you now say, "Hey, can we contact you directly? I have a question, like to chat, anything." Of course, drop me an email, NB@NB-Networks.com. Always looking forward to hearing from you there.
And when you say, "Do you have anything which is live where we can talk to you live?" Yes, of course. When you go to expert.nb-networks.com, just sign in there. Very important is here, as soon as you sign in there, you only receive one email every Wednesday morning. It's 100% content, ad-free guarantee. That's the important bit here. There is no advertising whatsoever. Full access to all the podcasts, videocasts, anything, and of course, all the special events that we have.
It's only available via the leadership letter. There is no other channel where we are going to advertise with this. So be sure that you really tune in, because as soon as you don't know that it's happening, you can't join. And by the way, it's free. And by the way, when you now say, "Hey, we need a live session, but a private one. Is that possible?" Of course, just drop me an email, and then we take it from there so we find a private session for you.
Of course, you can also connect with me. Feel free to connect with me on LinkedIn. Don't do the follow thing. Connect properly. I'm looking forward to being connected with you there. Then of course, on top of that, you can follow me on Instagram, and of course, you can leave a like on Facebook, or you just subscribe to my channel on YouTube. All of that is available.
The most important thing, however, is always the last bit that I say. Apply, apply, apply what you learn in this podcast and videocast, because only when you apply, you will see the positive aspect that you obviously want to see in your organisation. I wish you all the best doing so. If you now have anything where you think, "Hey, I'd like to discuss," just drop me a line. Very important. And answer is guaranteed within 24 hours or less. So I'm looking forward to hearing from you there.
At the end of this podcast, as well as at the end of this videocast, there's only one thing left for me to say. Thank you very much for your time.