The AI challenge – adjust your performance expectations and measurements now!

Robert Shore: Hello, and welcome to the Brightmine podcast, formerly known as the
XpertHR podcast. Brightmine is a leading provider of people data, analytics and insight, offering employment law expertise, comprehensive HR resources and reward data to meet every HR and organisational challenge and opportunity. You can find us any time of the day or night at https://www.brightmine.com.

Hello everyone. My name is Robert Shore, and we’re going to be talking about AI and the way that it may be increasing workloads rather than reducing them, and what HR needs to be aware of and what it can do about it.

To discuss this thorny subject, I’m delighted to be joined by Jeremy Hollander. Hello Jeremy.

Jeremy Hollander: Hello.

Robert Shore: And possibly I’m going to get you to introduce yourself here, I think,
because it’s too complicated for me. I can’t handle all the technical stuff!

Jeremy Hollander: Happy to. Thanks for having me. Just as a brief background, I spent the last 20 years in enterprise technology, starting as a strategy consultant for one of the big tech companies out there called Cisco, and then I worked with their Fortune 500 customers. And then I went way downstream and moved to working with startups and eventually also founding two venture-backed companies in the HR
and AI space.

And I think what that gave me is that journey sort of gave me a close view of what actually happens when AI lands inside an organisation. And what I keep seeing is the gap between the promise and the reality is not always about the technology itself, but it’s also about the way that companies think about work, and it’s about how they measure it, how they reward it and also how they design it.

Robert Shore: There we go. I think that introduces our theme today very well. And as you’ve outlined briefly there, you are quite expert in this field. So we said we were going to begin our chat today by talking about a piece of research that’s been published in the Harvard Business Review, which says in short (this is the headline), “AI doesn’t reduce work; it intensifies it.” So researchers studied 200 employees for eight months and found that AI intensified work, did not reduce workload. Workers voluntarily took on more, worked longer hours and expanded their scope.

So first question, Jeremy, is did this surprise you, this finding, or confirm something you were already seeing?

Jeremy Hollander: I think it’s surprising. It’s surprising because you would think with AI you would spend less time working. To me that’s just…it’s common sense, right? And everyone knows this. You enter something in AI and you get a response within seconds and you’re done with the work that would otherwise take you half a day.

But I think what we’re seeing – and I just even talk about myself – what I’m seeing is I end up working on three, four, five different projects at any one time because I can. And I think it’s becoming more taxing. It’s just becoming more taxing. So I think it’s actually counterintuitive that in some ways AI is perhaps making us more productive but also making it harder for us, harder for employees.

Robert Shore: So is this on HR’s radar yet or is it still flying under, in your experience?

Jeremy Hollander: I think it’s still flying under. You know, if you look at ChatGPT it was introduced less than three-and-a-half years ago and that was really for the early adopters. And if you look at the enterprise, I think candidly up to even less than a year ago I would talk with enterprises and they would say, “Look, we don’t want our employees to use ChatGPT at work. We don’t want them to use Claude at work.” And I think there’s been a shift. There has been a shift of about six or nine months where people say, “We have no choice anymore. If we do not integrate AI tools we’re going to have trouble competing with other companies.”

So I think there has been a shift over the past, say, certainly couple of months.

Robert Shore: You know, certainly this technology also makes people feel more capable in many ways. You know, you can…certainly the fear of the blank page goes away. You can just type something into one of these various engines that are available, you get an answer back, it can format something for you, give you, you know, structured text straight away. You feel very capable as a result. What’s the problem with that?

Jeremy Hollander: Wow. I mean, it’s almost like giving people superpowers, right? And you know, if you look back you’ll say, “Look, without AI tools everyone is as good as they are. They need to learn. They need to learn anything new that’s being asked of them. If they do not know today, they need to learn in order to complete a task.” And I think the challenge now is that with some of those AI tools – and I really think the challenge is – anyone can do anything. And it is borderline shocking. You could ask an intern to do something that otherwise would require someone with 10, 15 or 20 years of experience to do. Why? Because the intern has access to essentially the same type of background and data or “experience” (with massive quotes) that someone with 20 years’ experience would have. So I think that really shifts the question to, “Okay, we now…anyone can do anything. So what does that mean?” Can you just ask any employee to do anything? And I think that’s a question for HR.

Robert Shore: Yes. I suppose one question I’d ask there is, do you think when the intern with no experience does it, they do it as well as the person with 20 years’ experience could do it?

Jeremy Hollander: It’s hard to say, “Yes, they do it as well,” because then I think you’d minimise the value of experience. And I would argue that they don’t do it as well as someone with real field experience because there is something to me that’s incredibly valuable – it comes with experience, right? It’s judgement. It’s looking at “What am I trying to do?” You know, if you look back at maybe 20+ years ago, you know, there was this notion of “Do you know how to Google something?” And people would say, “Well what do you mean?” Well it’s one thing to say, “I need to search something,” but do you know how to really Google something, the keywords to use and how to look for data? And I think there’s something similar now with AI tools, is, “Do you know how to look for the data?” And when you get data back, you can get so much data back so quickly, now you need to make a judgement call and say, “What do I take?” Because you can no longer look at this as a productivity problem because it’s not a productivity problem. Anyone can have access to as much data as they want at their fingertips. It’s more about “Okay, I can have so much data in front of me. What do I do with this data?” You know, what’s right, what’s wrong, what do you extract from it? It used to be that you’d say, “Well, you know what? If I serve or if I hand out this massive essay, massive thesis, it’s going to be incredible.” I think people value now really short summaries. Because all these AI tools can create so much data but now you really want, “Give me the gist of it. What’s your thought based on everything you’ve seen?”

Robert Shore: And that, as you say, does take experience and wisdom of some kind, doesn’t it, that is accumulated across working life. But I think one of the things that comes out of the Harvard Business Review piece is that people are sort of voluntarily working harder as a result. They’re taking on more because of these superpowers. Is that a problem? They’re not complaining, apparently. So should HR be doing anything about that? Is this an issue for HR?

Jeremy Hollander: They’re not complaining yet. And I think, look, we just started, right? I think this AI has really been integrated, or is getting integrated and readopted in enterprises over the past year. It’s moving incredibly quickly. And the way people look at this is they say, “Okay, this is just an adoption problem, right? We get the right tools, does it pass our security test and policies and whatnot?” But we’re not really thinking about the other side, we call it the human side. You know, we’re essentially giving all our employees access to this incredibly knowledgeable tool which almost is like the inventor of – I don’t know, maybe I’m dating myself here – AltaVista, right? Back in the nineties. And Yahoo. And then Google. But they have access to a world of information. What does that mean? There are so many questions.

I think, you know, one question for example is, “What does this mean from a performance management perspective?” Another question is, you know, “Are people going to get tired?” Tired just of context-switching so often, so much, and not having potentially a single break during the day. Why? Because every moment that you have to…as you wait for one prompt to complete, you could start another prompt, “another project”.

And I think there are a lot of questions that HR needs to answer. And I think we will learn. We will learn in this year, the coming year in 2026, what are some of the downsides. Or maybe not downsides; maybe the things that companies, enterprises need to think about from a human perspective when adopting AI. And I think not much research has been done on this yet.

Robert Shore: So obviously AI is different from the introduction of email in the nineties, which obviously changed things a lot, and then the smartphone, which really revolutionised…well, things way beyond the workplace. You mentioned there about what we measure, really, in terms of work. Are we talking about measuring things differently now or, you know, in terms of productivity and output how should HR be thinking about that?

Jeremy Hollander: I think measuring has always been hard. It’s hard because it’s…in a way you say, “Look, we’ll measure,” and it’s very scientific. It’s a number. That’s wonderful, but what you measure is important too. It’s just not how it is measured but what are you measuring? And I think historically people will look at this and say, “Look, we look at productivity, we look at output,” right? And yes, that’s what worked historically but now output has become irrelevant because you can output so much work with AI in a fraction of a second, that I think it is difficult to just look at pure output.

I think a better way to look at this is to look at business outcome. Which really should always be looked at, right? And so, you know, if you look at engineering, you know, I don’t think it was ever about, “Let’s look at how much code an engineer writes per minute.” Or per hour or per day. Because some code is harder to write than others. Some problems are more difficult to solve than others. What is the outcome of the code that they write? What is the business outcome? And I think the same thing is in marketing, right? It’s not about, “How much copy can you generate in a day?” It’s, “Did you generate valuable copy that brought on more customers?” That is the question. So really it comes down to business outcome.

So to me it’s less about productivity and about pure output, but it is about what are the metrics or the KPIs at the end of the day that truly impact the business? Let’s figure those out and then bring this back to employees and see how do they perform on those metrics, as opposed to, “Oh, did they just produce a massive amount of text?” because that is today no longer relevant.

Robert Shore: Yeah. ‘Cause I mean, as you were saying earlier with sort of somebody who comes in a space – it could be an intern – they’re able to produce an enormous amount of work. Who then reads it is another matter, and how it’s assessed. I mean, I’m aware that sometimes I have…I mean, obviously things I’ve done for my personal interest, you know, produced things without necessarily knowing anything about it at the beginning, and I don’t necessarily know anything more about it at the end either. But I have produced something that apparently contains the gist of what it’s supposed to be about. I mean, there’s quite a lot of work you can do now without knowing anything at the beginning, or indeed anything more at the end. So I suppose that was the old thing, that you’d begin possibly by not knowing much but then you do a certain amount of research and you have to learn, whereas of course AI means that you can sort of fly blind on some of that stuff now.

Jeremy Hollander: You can. And by the way, you mentioned an intern can produce work very quickly, a lot of work. And here’s what happens. An intern – not an intern, anyone – can produce an enormous amount of work today with AI, and your question is, “Who reads this afterwards?” The AI produces it. At the other end, on the receiver, the receiver will say, “This is just too long. Let me put this into an AI to summarise it for me,” whereas the person, the sender, could say, “Let me write this in three bullet points,” and the receiver would be just as happy to get this in three bullet points as well. So you almost have the inverse happening now where you have one person say, “I need to produce something very long,” and on the other hand say, “I received something really long. What am I going to do with it? Let me put this in an AI to say, ‘Summarise it for me again.’” So people think that you’re creating so much more work but in reality, I think, today people have less and less patience to read. Because it’s so easy to create content that you just want to have the gist of it every time. And to me again it comes back to business outcome, right? What it is you want to say? Say it in a short way, say it in a simple way, and then the person will be happy. This is what people are looking for today. I think we go back to the basics.

Robert Shore: Yeah. And I mean, is there a sort of formula for doing it that way? Is it just about, you know, continually feeding it into some sort of AI bot and keep saying to it, “Can you put this into, you know, five points and then three points and then…ooh, could you possibly get it down to just the one bullet point?”

Jeremy Hollander: And I think that’s what I said earlier. I think it goes back to just raw human intellect. And at the end of the day, I think the most valuable thing that someone can bring to the table is not going to be, “How quick can I type?” but it is, “What do I understand or have the ability to learn something new, something quickly, and then make a judgement call?” Because I can go there and type anything now and read about it and send an email about it. But do I understand the response? Do I understand what AI is telling me? And if I do, that’s called raw intellect. I can go so much further today.

So I think in many ways – maybe going back to your question earlier – AI is helping a lot, sure, the call it newcomers to the workforce. But I think it’s helping even more the high performers to say, “Look, I can do anything now. I’m going to work harder…” – which potentially is a problem – “…but I have anything and everything at my fingertips. I can learn anything, anything overnight.” And I think that’s very, very powerful.
But from an HR perspective as well you must be aware, you know, this can be incredibly taxing. And you need to know – and I don’t have the answer yet because we haven’t seen this happen truly for enough time – but changes will need to take place over the next year or two because we’re going to otherwise have, I think, challenges in the workforce. It’s going to be very tough for employees to handle.

Robert Shore: Performance review, then. Performance review problem, we’re calling it. I think you came up with a neat little challenge, which is if a salesperson closes twice as many deals because AI is writing their outreach, do they deserve twice the bonus? What is the answer to that?

Jeremy Hollander: I’m not the chief revenue officer! So this is my opinion. But good for them. They should absolutely get 2x because they’re leveraging tools in the right way, in the way that they were meant to be used. And if it’s sustainable, good for them.

Now, I think that the baseline’s changing, right? People are…if you go back to your example, right, of salespeople, on the prospect side, you know that people are using AI tools now. So the baseline of, “What does it take to close a deal?” is getting harder and harder. So I think the goalposts are moving over time as well. But I have absolutely nothing against people leveraging AI to do their job. I think they should. I think it is wholly wrong for companies to say, “We will not use AI.” I think these companies will be left behind. And I know a lot of companies who are now adding metrics and telling people, their employees, and saying, “You must use AI. If you do not use AI, you have no place here.”

And I think the question to your question is, “How do you measure this?” And I’ve seen simple things of saying, “Well, how often do you log into ChatGPT or Gemini or Claude? That’s great.” And I’ve seen more detailed ways of looking at this and say, “Well how many tokens have you used?” And this is becoming quite detailed. Is there a right answer to that? I’m not so sure yet. We have to figure it out. But no question, you must use AI at work. But to use AI, companies, employers, have to enable their employees to use it. They have to get access to the right tooling. They have to train their employees. They have to have their employees adopt it. But the baseline, the goalpost is moving, right? So this one employee may be doing 2x in revenue this quarter, and hopefully the next quarter more employees will be doing the same, but then the goalpost is moving, and so that 2x will go down to 1.5 and 1x because the market is catching up with that.

Robert Shore: So there’s a piece of research that’s just been published, done by YouGov, about AI adoption in the workplace in the UK, which is that actually a lot of it is self-directed. So you were talking about obviously employers facilitating the adoption of tools, but what do you think about this sort of phenomenon that actually an awful lot of what people are doing in the workplace is actually according to their own lights, that essentially they are, you know, messing around with stuff in a constructive way, trying to find what’s best for them, but a lot of it is done without a sort of guiding hand from the employer. Is this a good thing? Is this a problem?

Jeremy Hollander: The first question I would ask is, “Why are the employers not helping them?”

Robert Shore: I mean, it may be because the employers don’t know better than the
employees…because it’s so new, isn’t it?

Jeremy Hollander: Look, I think as long as the employers don’t prohibit the employees for using AI tools, as was the case as early as, you know, a year ago, where they said, “Look, you are not allowed to use AI tools until they move forward.” I think employees should trial, test and use AI as much as possible but also be aware of what they’re doing, right? Depending…there are many ways to use AI, and if you…as long as you are aware of how you use it and what the consequences could be, and again you remain in the loop, right – there’s this notion of “human in the loop” with AI where you say, “Look, AI, go and do this. Give me the output. I want to view it before taking action and for example, before pressing send on an email.” And I think that’s great. Where it becomes a bit more tricky is where you have agents doing the work for you, where you say, “Here’s my email. Here’s my username and password. Answer my emails. I’m going to go to the beach and when I come back I want my work done.” That’s trickier because I don’t quite think we’re there yet. And if you do that, obviously you’ll be putting not just your career at risk but also the employer’s perception put potentially also at risk.

So I think employers will catch up and certainly need to help employees adopt, train employees on the way. But as long as employees can trial and test AI in a safe environment – and by safe environment, to me it also means the employers are okay with mistakes; that’s a safe environment – then I think it’s okay, I think it’s fine and they absolutely should.

Robert Shore: I mean, part of the sort of dream of the future is yes, you get agents to do the work for you and then it’s an agent that reads the output and then…I mean, just when I receive Gmail now, which is not my work email, you know, it gives me a summary straight away. And I’m quite annoyed, actually. And I’m thinking, “Why are you reading my email? Get out of there!” Or at least, “Can you respond to it at the very least? If you’re going to read it, then can you get on with doing it?” But anyway, that’s a distraction. Let’s not go down that route! We were talking about performance frameworks and whether these then are going to need to be rethought fundamentally in the light of AI as a tool, which changes, you know, the metrics of so many things. What is the future of that sort of performance review framework?

Jeremy Hollander: It’s a tough question.

Robert Shore: Good! Would you like to type it into AI quickly to answer it?! I would have to. I know you don’t have to!

Jeremy Hollander: No, I don’t think that there is a right or wrong answer yet because we really are still just getting started with adopting AI in enterprises, right? I mean, consumers, billions of consumers are using AI. I think in the enterprise they’re just getting started. I think a lot of mistakes will be done. For example, do you really want to count the number of tokens, you know, to see if your employee is doing a good enough job with AI? I don’t know. To me it seems a bit like a vanity metric. Do you want to see how often someone logs into AI? I don’t know. I think it goes back to the business fundamentals. I think it goes back to, “What are the business outcomes?” And you know what? If you choose to not use AI today, okay. But you probably should know as an employee that in the next three, six, nine, 12 months, if you don’t keep up with AI, you will most likely fall behind other employees. And so from a performance perspective, if we’re looking at pure business outcomes – which may be cost, which may be revenue etc – you will not be able to keep up with other employees. And this will almost be a forcing function for you to join into AI.

So bottom line for me, go back to the fundamentals in performance management. It is…I think it has become so complex, you should attempt to tie your employees’ performance to pure business outcomes, see how they’re doing and be honest with people and say, “Look, we’re learning. We’re learning along the way. We’re not going to make drastic changes overnight. We’re doing what we can. But we know that we need to do more. We need to change the way that we measure performance. We will probably make mistakes. But we will learn and we’ll do it as quickly as we can. And we’ll also do our best to do it right by the employees.” Because when you learn, you are about to make mistakes. And I think employers who are honest should also attempt to do what’s right not just for them but also for the employees.

Robert Shore: And actually, just very quickly, from what you’ve been seeing, are employers looking at sort of token use and logins? Is that being used as a metric that you’re aware of?

Jeremy Hollander: It is. It is being used. It is being used and you can track it. You can connect to the different LLMs and get access and see how often does Robert log into ChatGPT or how many tokens has Robert used over the past day or the past week, and then let’s compare it with someone else who is in a role similar to Robert’s role, and let’s see how they’re doing. I hope and I assume that those metrics are a tiny, tiny slice of the bigger pie because they absolutely would not give you a big picture of what is actually happening. It’s, I think, to me, just one of the many metrics. Arguably a vanity metric, but we haven’t seen enough yet. Maybe employers are just using this as a forcing function to get their employees to use AI more and more, and then the use of things like how many tokens, how often have you logged in, will slowly fade away and make place for more important metrics.

Robert Shore: Let’s come up with some takeaways for HR. If the working week begins on a Monday morning, what is the one thing a HR leader should be doing first thing on that Monday morning or when they begin their work week?

Jeremy Hollander: Train your employees on using AI. I would say forget everything else. Performance, forget everything else. I think it’s, “How can you enable your employees to use AI to improve business outcomes at your company?” It starts with training. It starts with helping people understand. It starts with getting the right tools, the right processes in place. Because you cannot expect every employee to know what to do, where to do it, how to do it. It is not an innate thing for everyone. So train them. And once you’ve trained them and you’ve opened that can and said, “Okay, all of you now have access to AI. Here’s training as well. You have questions? Come.” You really make it an open environment, an environment where people feel comfortable trying things. This is cultural at this point, but trying things. And also making mistakes.

You will learn from this. And then you can go and look at performance and say, “Okay, what have you learned? How can we shift our thinking from performances A, B and C to performances down to something different?” because the goalposts have changed
entirely with AI.

Robert Shore: So I think what you’re pointing to here is this is the moment for a fundamental reset in many ways, and that actually this has to be installed in this period. This is really a critical thing for organisations to be doing right now in these months. Is that your sense of this?

Jeremy Hollander: I think “reset” is a big word. When I think “reset”, I’m thinking the
button…

Robert Shore: It’s just five letters. Five letters, Jeremy!

Jeremy Hollander: I’m thinking about the reset button on a desktop computer back in the nineties. You know, you press that button and everything just blows up. Or CtrlAltDel. I don’t think you want to obviously reset things overnight. I think that there’s an approach to everything. Look, again people will make mistakes and so you want to take a measured approach.

So to me, “resetting” is saying not just, “Let’s throw everything out,” let’s look at it, let’s come up with a different approach but let’s take a measured approach. And this is a journey. It’s a journey for the company as it is a journey for AI as well. AI is moving so quickly, no one knows where we’re going to be in three months, let alone six, nine or even 12 months. And so knowing, for employers to know that the field is moving so quickly, as long as they are aware that this is moving quickly and they will need to make changes over time, and they message this, they communicate this to their employees as well, everyone will be on the same page, it will be a safe space, everyone will want to learn – “everyone” meaning both employers and employees. And I think they will then end up in a much, much better spot than they would otherwise.

Robert Shore: And if you are a HR leader about to undertake this kind of operation and you’re possibly not yourself that savvy about these things, how do you get the confidence to know that you’re doing the right thing, that you’re pointing employees in the right direction, that you’re providing the right training, the right encouragement? Is there a good website that gives you advice on this?

Jeremy Hollander: Use it. Use AI every day, day in, day out, more than you would expect your employees to do so. And I don’t think that any amount of or any number of websites can help you figure this out. The best way is to use it, to try it and to learn from it. Because I don’t think you can expect anyone to use it until, and/or unless you use it yourself. Start with yourself, look at the opportunities, try it out, and then tell your employees, “Look, we’re looking at X number of tools. Try that as well. Let us know what you think. We will learn from this and we will get better.” And I’m saying this because the field is moving so quickly it is hard to keep track. And the best way to do this is to try and to use and to try again.

Robert Shore: So you’ve got to get your hands dirty yourself?

Jeremy Hollander: Yes! Yes, you…not potentially. Definitely. You must. This is the only
way to use this and to get used to it, and to help others adopt it too.

Robert Shore: What does “good” look like if HR gets this right? How are we going to
measure our success as HR, our own success in getting it right?

Jeremy Hollander: I wouldn’t put a metric on this. I would look at this as, “Did we fail? Or did we learn from our mistakes?” And if we did, then I think we succeeded because there is no easy way to say, “Oh, this is a success,” because a success today could be a failure in three months. The best way for HR to look at this is simply to say, “Look, we don’t know what we don’t know. We need to learn and we need to create a culture and an environment at the company that is open enough for people to learn, to try, to make measured mistakes over time.” Because what that says to your employees is, “Look, we don’t know where we’re going with AI but we want you to feel comfortable trying because we know that this is the path forward.” And what that’s going to tell employees is that it’s going to let them know that this is a great environment, a great company to join. They will feel comfortable trying things out. It will be good for you, it will be good for your employees and it will be good as well for the way that people perceive you as a company, and say, “Wow, that company really is open to any and all ideas.” And again, do this in a measured manner. But that to me is what success will look like.

Robert Shore: There’s a degree of psychological safety involved in this.

Jeremy Hollander: Yes.

Robert Shore: Jeremy, that is wonderful. Thank you so much for your time today. As ever, of course, we have some supporting material in the show notes for this episode. So I’ll thank our guest, Jeremy Hollander, again.

Jeremy Hollander: Thank you, Robert. It was a pleasure.

Robert Shore: At that, I will just say, until next time.