There is growing consensus that AI is eroding the experience pathways organizations depend on to develop human judgment. Entry-level roles are shrinking, hands-on work is being automated, and fewer people are building the experience their businesses need. This is a well-documented challenge. What remains far less clear is what the talent strategy response should look like. Most skills systems today are built to track whether someone can execute a task, not how well they make decisions under ambiguity. If judgment is becoming the most important capability in the organization, recruiting, learning, and assessment all need recalibrating around it.
So what does that recalibration look like in practice?
My guest this week is Craig Friedman, author of Enterprise Skills Unlocked and Talent Transformation Leader at St. Charles Consulting Group. In our conversation, he explains why current skills infrastructure misses the capability that matters most, what work looks like when execution is automated, and how organizations can start building judgment into their talent strategy.
In the interview, we discuss:
- What does enterprise AI adoption currently look like
- The long-term consequences of cutting entry-level jobs
- What does work look like when the automatable tasks are taken away?
- Why judgment is more important as a skill than ever before
- The difference between judgment and critical thinking
- Where does judgment fit into the skills infrastructure?
- How can TA teams hire for judgment?
- What does the future look like?
Follow this podcast on Apple Podcasts.
Follow this podcast on Spotify.
Key takeaways
- Craig Friedman argues that concern about AI eroding judgment obscures a deeper problem; most skills systems were never designed to develop or measure judgment in the first place.
- Entry-level roles are shrinking and hands-on work is being automated, cutting off the experience pathways organizations depend on to develop human judgment.
- Current skills infrastructure tracks whether someone can execute a task, not how well they make decisions under ambiguity.
- If judgment is becoming the organization’s most important capability, recruiting, learning, and assessment all need recalibrating around it.
Transcript
Matt Alder 0:00
There’s growing alarm about AI eroding the judgement that develops through hands-on experience. The concern is very justified, but it also obscures a deeper problem. The skills systems most organisations rely on were never designed to develop or measure judgement in the first place. What does a talent system built for judgement really look like? Keep listening to find out.
Advert 0:26
Support for this podcast comes from Paradox. You know, I’ve recently heard some crazy success stories when it comes to hiring with AI. FedEx is sending offers to candidates within 10 minutes. General Motors saved $2 million in recruiting costs in a year. Chipotle reduced time to hire from 12 days to four. And 7-Eleven is saving their store leaders 40,000 hours per week. The craziest thing is all of these companies did it by leveraging the same technology. Paradox.
Acquired by Workday in 2025, Paradox pioneered conversational hiring. Whether acting as your ATS or automating the hiring process on top of your existing HCM, Paradox streamlines the entire experience. And if you’re a Workday customer, Paradox now operates seamlessly within their talent acquisition suite as the Workday Paradox candidate experience agent.
Driven by a 24/7 AI assistant, Paradox can handle up to 95% of the hiring process for deskless hiring teams, or just automate specific, time-consuming tasks like screening, interview scheduling, and onboarding to allow recruiters to focus on recruiting. Paradox continues to help hundreds of the world’s top employers simplify hiring and save money, while creating great candidate experiences. Spend more time with people, not software. You can find out more by going to paradox.ai. That’s paradox.ai.
Matt Alder 2:28
Welcome to episode 816 of Recruiting Future with me, Matt Alder. There’s a growing consensus that AI is eroding the experience pathways organisations depend on to develop human judgement. Entry-level roles are shrinking, hands-on work is being automated, and fewer people are building the experience their business needs. This is a well-documented challenge. What remains far less clear is what the talent strategy response should look like.
Most skill systems today are built to track whether someone can execute a task, not how well they make decisions under ambiguity. If judgement is becoming the most important capability in the organisation, recruiting, learning and assessment all need to recalibrate around it.
So what does that recalibration look like in practice? My guest this week is Craig Friedman, author of Enterprise Skills Unlocked and Talent Transformation Leader at St. Charles Consulting Group. In our conversation, Craig explains why current skills infrastructure misses the capability that matters most, what work looks like when execution is completely automated, and how organisations can start building judgement into their talent strategy.
Matt Alder 3:46
Hi, Craig, and welcome back to the podcast.
Craig Friedman 3:50
Hi. Thanks for having me again, Matt. Always a pleasure to talk to you.
Matt Alder 3:52
Could you just introduce yourself and tell everyone what you do?
Craig Friedman 3:56
Sure. I’m Craig Friedman. I’m the Skills and Talent Transformation Leader at St. Charles Consulting Group and author of the book Enterprise Skills Unlocked.
Matt Alder 4:04
Fantastic. So let’s talk about AI. Let’s talk about skills. So there’s a huge amount of noise around AI at the moment. Obviously, I think every single episode of the podcast, we discuss it now. From what you’re seeing, inside large organisations, how is it actually being used on a day-to-day basis?
Craig Friedman 4:23
Yeah, that’s a great question. There’s a pretty wide adoption gap. Like, it depends a lot on the company. Honestly, there’s quite a lot of companies that are only now sort of rolling out their enterprise version of AI that’s taken them some while to sort of settle on, train it, lock it down in a way that they’re comfortable with, and they’re only now starting to roll it out. There are some that are further along those journeys. In terms of like specific usages, again, it frequently depends on the group. Like you see sometimes more advanced adoption in some narrow applications. Like, I mean, obviously tech firms are using it quite a bit for programming. Yeah. You see quite a bit sort of in the chatbot and support automation area, but the sort of success and satisfaction is uneven. You’re seeing more success, sort of high volume, low complexity kinds of queries, but rolling it out to more complicated areas is more mixed, you know, if you will. The general population, you probably see a lot, probably the most common application is like just drafting communications, emails, meeting summaries, initial draft of a report, and a search function, right? And then you have certain particular groups that are playing around, so like finance is playing around with AI analysis of spreadsheets and communications, and sales groups are playing around with quickly modifying and updating a PowerPoint deck that they have to deliver to multiple people that’s relatively similar. So can you quickly take the old deck and modify it for a new customer or a new department or something. So it depends on the group. L&D groups are playing around with AI role plays for teaching purposes, like in areas like customer service. So it’s really scattered.
Matt Alder 6:00
Yeah, no, absolutely. I mean, definitely, that definitely seems to be the case. We’ve been talking for a while about how AI will disrupt jobs. And it seems that a lot of the early disruption that we’re seeing is kind of firmly in the area of entry-level jobs. What are the long-term consequences for organisations around that?
Craig Friedman 6:20
Yeah, well, there’s probably a couple. So the first thing is, if you are hiring less at sort of the bottom of your organisational pyramid, then your pyramid starts to look a little bit more like a diamond, doesn’t it, right? You don’t have a lot of people at the entry level, and then it gets sort of thicker in the middle with middle managers and more experienced types, right? Which presents an obvious problem of how is that smaller group at the bottom going to eventually graduate to expand into the fatter middle, if you will. And so that’s an org design challenge. But on top of that, it’s actually even more complicated because the people who are left at the early career ranks, they aren’t getting as much exposure to the work because the AI is automating a lot of what they’re doing. And so those early three, four years that they would spend drafting various kinds of documents and having their supervisors review it and correct it and explain to them why it’s wrong and sort of pass on their wisdom for, you know, how to do this well, they’re not getting as much of that. So yeah, even the ones who are left aren’t developing the judgement to become the experts that they need to become.
Matt Alder 7:30
Yeah, no, absolutely, that makes a lot of sense and the kind of the real model of that is under threat, isn’t it? It seems to be kind of an existential risk if people can’t develop the expertise they need by learning on the job.
Craig Friedman 7:39
If you think about it, it’s never been more important, right? Because AI is wonderful at generating content, but it takes quite a bit of work to get it to generate sort of expert content. It needs a lot of review, especially in the early phases. It makes all kinds of mistakes. I mean, hallucinating is the most obvious one, but it makes logical errors. It leaves things out sometimes. It doesn’t always stay on point. You have to really babysit it until you can train it to produce the kinds of documents that you want, which takes expertise and deep technical expertise, because what it spits out is very fluid. So you have to read it carefully sometimes to make sure it’s accurate. And that’s true whether it’s code or text content. So we’ve never needed critical thinking and expert domain judgement more than we do now with these AI tools, because the sort of manual work of just producing content is being automated. And what we really need is the refinement of an expert eye.
Matt Alder 8:42
You sort of mentioned critical thinking there as being a kind of a real sort of essential skill. What else is kind of essential alongside that? What other things do employers need to be thinking about here?
Craig Friedman 8:49
So first of all, the critical thinking, there’s two bits to that, right? The one piece is the sort of process of thinking clearly, right? Knowing how to analyse things carefully and not make mistakes along the way and sort of the discipline of thinking thoroughly about something. But the second bit, which is sort of the really important one that’s being overlooked, is the judgement, right? And by judgement, I mean the ability to make decisions, particularly under ambiguous kinds of situations, right? Where maybe I don’t have all the information I need or the right answer. There’s more than one right answer. It sort of depends on the circumstances, or I have to sort of decide whether I’m trading off between two different goals that are both kind of equally important. So at the end of the day, it becomes a judgement call. There’s no clear right answer. And you rely on your expertise to make those kinds of judgements.
And again, we talked about how AI is sort of hollowing out the expertise development part of the process in a couple of different ways. So you’ve got both of those elements. And so you’ve got a combination of the process of critical thinking and then sort of the raw judgement capacity that comes from the accumulation of experience over time. And not just any experience. Typically, it’s experience that is guided by a supervisor to help you develop over time.
Matt Alder 10:11
Absolutely. And I mean, you know, we talk about skills a lot. I mean, where does judgement fit into a kind of a skills-based world?
Craig Friedman 10:19
Yeah, that’s a great question, right? Well, so from a certain point of view, judgement is a skill in and of itself, or at the very least, you could think of it as a sort of the top level of a number of different skills, you know, where you can sort of develop materials, apply concepts, analyse them, but the top level would be judgement, if you want to think of it that way.
And so we have all this, we have all this infrastructure today where we teach skills and we track skills, right? We’ve got learning courses that teach various skills and assessments that try to assess, you know, what your level of skill is, and we can put it in a skills taxonomy and put it on your profile and all that stuff. But most of that system is not optimised for judgement. The nature of the way we teach and capture and talk about skills is largely sort of procedural skills in the sense of they’re tracking whether you can do something, right? Not necessarily how well you can do it, but just that you can execute the task, right? Because that’s sort of the most critical for, can I get you staffed? Can I get you busy doing something, right? This idea of, you know, judgement and critical thinking is largely at a different level. It’s, you know, with what expertise can I get this task done? Can I do it better somehow than I used to do it or than more junior people can do it? And the system isn’t optimised currently for that kind of refinement. And so what we would need to do is to tweak the nature of our training programmes and our assessment tools and sort of the whole talent system around skills so that it’s optimised to develop judgement, right? So it knows what we’re looking for and what the definition of better judgements are, and we know how to teach it, we know how to see it when it’s there, we know how to capture it, we know how to apply it to staffing decisions and career decisions and things like that.
Matt Alder 12:05
And how, I suppose the thing with judgement is it’s always kind of, it’s always sort of taken for granted that it’s something that develops with experience over time. You know, obviously you’ve sort of laid out some of the things that companies can do there. But how are companies going to adapt to, you know, this kind of whole new world where, you know, they have to think about these types of skills very differently to they have in the past?
Craig Friedman 12:28
I don’t necessarily think it is like a massive overhaul. I think it’s a fine tuning, right? So, for example, I don’t think it would be too difficult for a company to talk to its experts and start with what do we think are sort of the most critical judgements that people make on a daily basis, right? Just prioritise them and then start to dig into with those experts. We want to be able to find people who are better at making these judgements than others. So help us define sort of a skills definition and the level of sort of proficiencies or competencies, you know, at higher and higher level of ability to perform this, right? So that we could select for it in a recruiting function. We know what we’re looking for, or we could select for it internally when we’re staffing. If we’re teaching someone we would be able to tell if they’ve advanced. If we were evaluating someone for promotion we would be able to tell who’s exercising the higher level judgement. But that sort of all begins with at first getting your experts to help nail it down and define it and find tools that can then measure it when they see it, right? Regardless of whether that’s, you know, starting out with sort of a rubric for a personal evaluator to try to figure out and eventually graduates into actual like sort of assessment instruments. Like maybe you go through a simulation and you’re asked to demonstrate judgements at sort of ever higher level of complexity and ambiguity and the ones who sort of come through having demonstrated that they can handle the more complicated situations get a better score, right? So like eventually we’ll get there, but we probably will start with experts sort of defining how they know it when they see it and trying to share that with others and, you know, making a standard out of that.
Matt Alder 14:05
From a practical perspective, because these are things that are all going to take time, is there anything that talent acquisition leaders or HR leaders can be doing right now to start to address this?
Craig Friedman 14:16
Yeah, I mean, without a doubt. Look, let me put it to you this way. If you think about the impact that AI is having on the way our work is designed, right? And if you remember the old fashioned process flows, you remember those like they had little rectangles that like we’re doing something and then you’d have these little diamonds where it’s like oh, if the client is creditworthy go in this direction and offer them an application, if not go in this direction and give them a denial, right?
So AI is largely automating all the rectangles, right? And so what that leaves for work is just a sequence of judgements, right? It’s here’s a question I have, here AI go analyse it, and then you make a judgement and then you ask it the next question in the process, right?
So we have to sort of understand what the sequence of those judgements are going to be. What does the work look like once all the sort of generation tasks are kind of automated? And obviously, there’s always a place where we have to verify the output of what AI is doing in terms of human-in-the-loop oversight. But even beyond that is to try to understand what the work even looks like, right? And then once you understand what the work looks like and what’s the nature of the judgements and what does the task look like when we’re just proceeding, you know, in some ways, this is kind of the work that leaders do at the top, like because when you progress up to the top, everybody sort of presents you with challenges and options and you just make a decision and you move on.
We’re just now sort of doing that at a much greater scale. That style of work is being brought down in the organisation because a lot of the sort of manual generation work is getting automated. So now we’re all little mini leaders of our many areas making decisions and we ask AI to tee up the options and the pros and cons and we decide.
So if you start with that idea of understanding the work and like, what are the critical judgements that we have to focus on? Then I think you start to make sort of targeted investments, you know, so like from a recruiting perspective, what are the skills I’m looking for in the marketplace that are sort of highest value that we’re looking at first and what might be second, right? For L&D groups, it might be, how do I prioritise my learning investments around sort of the really critical, either high value or high risk judgements that we know we need to develop and so forth.
Matt Alder 16:22
Absolutely. And as a final question, it’s obviously a very disruptive time. What does the future look like? How do you think things are going to develop around this kind of thinking over the next sort of two or three years?
Craig Friedman 16:33
So imagine a world a little bit forward, right? Where we’ve, you know, let’s say developed a lot of AI enabled learning simulations that are optimised to teach judgement, right? And so they’re replacing some of what is lost, right? They’re giving you simulated experiences, work experiences to try to replace a little bit of what’s being automated.
And then we add to that, you know, a talent marketplace. But instead of the talent marketplace just teeing up, do they have the skills to work on this assignment? It also comes with a metric we’ve developed that tags the sort of developmental capacity of this assignment. Like what skills would it develop, right? And so when we’re matching people to opportunities, we’re considering both, you know, not just, you know, do they have the skills to do the job or could they develop them here and how, what’s its developmental ability, right?
And that information is fed into dashboards that help make decisions so that we’re trying deliberately to develop the judgements that sort of are being lost, right? And so some of the productivity we’re getting out of AI, we’re reinvesting into this judgement development.
And then imagine a system where, because we’re now developing skills through multiple different channels and acknowledging the skills that you develop, not just on the job through performance reviews, but also in learning completions and developmental gigs and feedback you get from experts and external credentials. And so now we have all these different capability signals coming into the system, right? You know, the learning might say that Matt is at proficiency level two here, but the performance review says he’s only at one. Right? And then he’s got this external credential, blah, blah, blah. How do I reconcile those different capability signals to come up with the most accurate view possible of what skill level Matt is at?
But once you have a system like that, and this is all possible with the technology we have, it’s just orchestrating it. But once I have really accurate information like that, we could do a number of really powerful things, right? So first of all, I can use it as the metric against which sort of recruiting and L&D activities are measured because, you know, to what extent are recruiting activities actually producing those skill levels that are the most important to us, either internal or external, right?
And the flip side of that is on the work, right? At the aggregate level, at the enterprise level. I can take a look at the people who are staffed doing various tasks and say, you know, on average, people who are P2 or teams that are heavy in P2 who perform this task tend to be correlated with fewer defects or fewer deficiency citations or more accurate projections or whatever, right? And until I can isolate sort of, it’s the transition between proficiency level two and proficiency level three that really is critical because that’s where we see the needle move most dramatically. And that then, you know, circles back into the critical skills analysis so that we know where to focus most of our efforts, right?
Matt Alder 19:36
Yeah, absolutely. It’s going to be a very interesting time. Craig, thank you very much for talking to me.
Craig Friedman 19:43
Oh, it’s been my pleasure.
Matt Alder 19:44
My thanks to Craig. You can follow this podcast on Apple Podcasts, on Spotify, or wherever you listen to your podcasts. You can search all the past episodes at recruitingfuture.com. On that site, you can also subscribe to our weekly newsletter, Recruiting Future Feast, and get the inside track on everything that’s coming up on the show. Thanks very much for listening. I’ll be back next time and I hope you’ll join me.
