Most organizations approaching AI are struggling and running pilot projects that go nowhere. The common assumption is that the technology itself is flawed, over hyped, or too complex. However, the employers that are succeeding with AI have discovered something different. The technology isn’t the problem, and the real barriers are human. Employee resistance, fear about job security, and the inertia of doing things the way they’ve always been done.
What makes the difference between AI projects that fail and those that transform how teams actually work?
My guest this week is Taylor Bradley, VP Talent Strategy & Success at Turing. In our conversation, Taylor shares how he built grassroots adoption in his team by starting with simple prompt libraries, the framework for deciding what should be automated, what should be augmented, and what should be left to humans, and why every AI project is really a human change management project in disguise..
In the interview, we discuss:
• Unlocking AI’s full potential
• The most significant challenges when implementing AI in HR and TA
• Why AI pilots fail
• AI projects are actually human change management projects
• The inertia of the status quo
• Talent use cases
• When to augment and when to automate
• Breaking down roles into tasks
• Surprising measures of success
• How HR and TA roles need to evolve
• Considering adverse impacts
• What will the future look like?
Follow this podcast on Apple Podcasts.
Follow this podcast on Spotify.
Transcript
00:00
Matt Alder
Recent research has shown that AI pilot projects have a high failure rate. While it’s easy to blame the technology, the main reasons for AI failure are actually very human. So what lessons can we learn from the successful AI implementations in Tanhr? Keep listening to find out. Support for this podcast comes from appcast. Appcast is changing the game when it comes to hiring. Using powerful tech, data driven insights and deep recruiting expertise, appcast helps employers find qualified candidates quickly and efficiently. In fact, companies using Appcast see a 50% drop in cost per hire and fill roles faster. Whether you’re building a brand or filling mission critical roles, appcast delivers results that move the needle. To learn more, go to Appcast IE. That’s Appcast IO.
01:05
Taylor Bradley
There’s been more of scientific discovery, more of technical advancement and material progress in your lifetime and mine than in all the ages of history.
01:19
Matt Alder
Hi there. Welcome to episode 738 of Recruiting Future with me, Matt Alder. Recruiting Future helps talent acquisition teams drive measurable impact by developing strategic capability in foresight, influence talent and technology. If you’re interested in finding out how your TA function measures up in these four critical areas, I’ve created the free Fit for the Future assessment. It’ll give you personalized insights to help you build strategic clarity and drive greater impact immediately. Just head over to Mataulder Me podcast to complete the assessment. It only takes a few minutes. This episode is about technology. Most organizations approaching AI are struggling and running pilot projects that are going nowhere. The common assumption is that the technology itself is flawed or overhyped or too complex. However, the employers that are succeeding with AI have discovered something different. The technology isn’t the problem.
02:28
Matt Alder
The real barriers are human employee resistance, fear about job security, and the inertia of doing things the way they’ve always been done. So what makes the difference between AI projects that fail and those that transform how teams actually work? My guest this week is Taylor Bradley, VP Talent Strategy and Success at Turing. In our conversation, Taylor shares how he built grassroots adoption in his team by starting with simple prompt libraries, the framework for deciding what should be automated, what should be augmented, and what should be left to humans, and why every AI project is really a human change management project in disguise. Hi Taylor, and welcome to the podcast.
03:13
Taylor Bradley
Thank you for having me here.
03:14
Matt Alder
It’s an absolute pleasure to have you on the show. Please, could you introduce yourself and tell everyone what you do?
03:21
Taylor Bradley
Absolutely. My name’s Taylor Bradley. I’m the Vice President of Talent Strategy and Success at Turing. So my role focuses on building AI native workforce strategies that help Both internal teams as well as our customers unlock the full potential of AI.
03:40
Matt Alder
Fantastic. And tell us a little bit about.
03:42
Matt Alder
The work that Turing does.
03:44
Taylor Bradley
Turing is an AI infrastructure company and we have two primary categories, and to put it in layman terms, one is AGI advancement. So we work with the world’s leading AI research labs on enabling their large language models to get to the next iteration of itself, to continuously improve the model that they work on. So that’s certainly one aspect of what we do. And then the other is Turing Intelligence. And this is where I often get heavily involved, especially when working with chros of our clients to implement AI is Turing Intelligence. So Turing Intelligence is about taking all of this lovely expertise that we’ve developed, working with the world’s leading AI research labs, and applying it practically to the workflows of our customers.
04:37
Matt Alder
Really interesting.
04:38
Matt Alder
And you kind of got one foot in AI, one fit in talent. You’re just the perfect person to ask this question to, which is, what do you see as the sort of the most significant challenges at the moment when it comes to this applying AI into HR and the talent function, just in terms of adoption and things like that?
04:56
Taylor Bradley
Well, it’s interesting. I just landed a little while ago from an off site we had in London with my team to take another iteration of how we’re implementing AI. And I was reading an article on the flight that MIT put together that 95% of generative AI projects fail. And what I said in a post describing our success that we had at the off site is we’re that 5%, and that was not easy to obtain is we had to come up with practical solutions with AI and then apply it. But in that journey, that sounds really easy. I came to realize that every AI project is a human change management project in disguise. And what I’d shared recently with Google was that the largest barrier to AI adoption is not the technology, it’s the inertia of the status quo.
05:56
Taylor Bradley
So your chros, your chief people officer and your people teams are essential to getting AI adopted in your organization because at its core, it’s a human change management process to drive adoption.
06:09
Matt Alder
Yeah, I think that makes perfect sense. And I kind of had a similar reaction to that article as well. It just sounded like people were kind of tinkering around the edges, trying to sort of just make existing things better rather than sort of rethinking everything in terms of what AI can do, just bring it to life for us a little bit. I mean, what are some of the sort of significant use cases that you have for AI in your sort of talent strategy function.
06:33
Taylor Bradley
It’s a great question and how I apply it is first a process that we’ve come up with on how to identify what are good AI use cases. So you need to train your staff and I’m talking about my people team on this journey is we’ve trained them to understand what is the benefits or potential benefits of gen AI with augmentation of workflows and automation of workflows. And then I’ve encouraged my team based on the knowledge that they have now is to think of how would I rebuild a role such as an HR business partner, a talent development partner, and use AI to do many of or most of those types of different tasks.
07:14
Taylor Bradley
So that requires us to look at a role like an HR business partner, break it down into its elemental task based level and then identify which of these tasks can be augmented by AI and which can be automated and then which of those tasks need to be still human in the loop. And the beauty of that is you have people that are on the front line doing the job every day, taking a high level or a decent amount of scrutiny of where AI should be practically applied in their role today. And that’s where we’ve come up with some really interesting use cases. And a practical one has been my people operations team came back to us and they said we mapped out our role.
08:00
Taylor Bradley
There’s many different parts of our role that still is critical be human in the loop, let’s say sensitive benefit questions. But we get a lot of questions around these different tasks. And when I say a lot of last year we finished the people operations getting around 55,000 tickets is what we ended the year with. And they were manually responding to every single one of those. So we looked at it and said, okay, we’re going to create essentially a chatbot. So nothing that is unusual for folks to implement, but we want to implement it in a way where it’s more than just responding to your ticket in a robotic theme where it’s obvious that it’s an AI.
08:45
Taylor Bradley
It’s not that we want it to fake being a human, but we want Alan is what we ended up naming it to be your assistant in whatever role you have. So you can ask it about any type of traditional people questions, but you can also ask it about the different tools and the projects you’re working on and we’ll give you feedback. And so we started to chip away at that and now Alan is responding to about 80% of our tickets. And we’ve Had a really interesting metric is we actually had an increase in tickets. And when we dug into it, were concerned that maybe Alan wasn’t responding effectively. But we’ve actually become victims of our own success is our ticket count has actually increased to around 70, 80,000 tickets for the year.
09:31
Taylor Bradley
It’s because people are finding Alan useful, so they tend to ask it even more questions.
09:37
Matt Alder
So it’s a measure of success in lots of ways then.
09:40
Taylor Bradley
Yes. And now my team, the people operations specialist, recently have come back and we’ve been very transparent about this journey so other people can learn from us and raise questions and concerns. Hey, Alan is starting to do so much of what I used to do. My day is actually getting a little bit lonely and I don’t have a lot to do. And there were concerns raised about how’s this going to impact my job. And so that was another opportunity for us to go back and say, hey, you know what? In a way, you’re right. You’re no longer a people operations specialist because that’s what Alan’s doing now. In fact, your role has just evolved. You’re now an AI engineer that has deep HR expertise.
10:24
Taylor Bradley
So your role is now, how do you continue to expand this use case, refine this use case, how do you reduce the amount of compute that we’re using? Because that is expensive. And so it’s really interesting to bring folks along on this journey.
10:39
Matt Alder
Absolutely. And when you’re breaking down those kind of things into those tasks, what kind of criteria do you use to decide what can be automated, what’s augmentation, and what should be left to humans? How do you make those kind of decisions?
10:54
Taylor Bradley
Yeah. So there’s two questions that I like to at least have the team answer is, if we automate something, what is the adverse impact if the automation gets it wrong? That is a level of scrutiny that we should apply before automating. Because when I think automate, I think I am putting the task completely in the hands of AI, where there may be a human in the loop, but only after the fact, where they qa, what is it doing right? And so that’s one piece that perhaps it’s better to augment with AI, with having a human in the loop, training the AI and fine tuning it and then scale it into automating a process. But that’s the first bit of it. And then the second is, what is the upside for the individual?
11:55
Taylor Bradley
Is automating this task saving us thousands of hours a year in one of our other use case, or is automating this task saving maybe an hour or so a year, and then it may not be worth the engineering investment because at the end of the day we have a people team budget and we need to be prudent with how we allocate that.
12:19
Matt Alder
You kind of mentioned right in the beginning that your kind of realization that this is really all about change management and humans and that kind of stuff. How do you kind of manage that? And I think also you sort of alluded there to perhaps the danger people feel about their jobs as well, in terms of how do you give that kind of psychological safety? How do you know, how do you kind of run that as a change management initiative?
12:42
Taylor Bradley
Yeah. So you have to start by building grassroots support. And that’s the beauty of it is many people, leaders are looking at this as some type of technological problem and they feel that they may be inadequate of not understanding deeply AI. But that’s not what they need to understand. They need to understand that this is a human change management process. And so simple steps. First, you don’t want to go in and say to someone, map out your role and tell us what AI can do. That’s not the first step. That’s only when you have people on board that they understand how their role will evolve. Our first step was very simple. It was getting everyone ChatGPT, Enterprise or Gemini, whichever an organization may choose, and giving them a prompt library.
13:33
Taylor Bradley
And were very intentional about these prompts that we wrote so people could understand what great looks like when prompting. But the prompts were focused on day to day things. An example I love to share is the Diplomatic Dispatch. We named all of the different prompts. The Diplomatic Dispatch is a prompt that you would use that if you were going to slack someone a sensitive message or something that you feel would maybe elicit some type of reaction. They may be hypersensitive to the news or the question that you’re going to share. You run it through the Diplomatic Dispatch and it essentially says, hey, I have to send this message and here’s my draft message to someone that may be hypersensitive or hyper reactive to this news.
14:20
Taylor Bradley
Help me put my message in a way that still has the core of my message that still sounds like me, but is in a way that is constructive and will mitigate the reaction and focus on improving the outcome. And at first people thought it was the silliest thing. They’re like, I don’t need a prompt to help me write. But then when we encourage people to use it and they started to use it they started to see the benefit of artificial intelligence, where it was only complimenting, not replacing their voice. And then once you start to get that ball rolling, it snowballs. And you get folks that are starting to turn into AI advocates. They understand its limitations, but also its benefits. And then that’s when you dive into the deeper topics.
15:01
Matt Alder
You mentioned that people’s roles are kind of already evolving from what they were doing before to kind of really sort of managing the AI and, you know, doing the parts that are still human, all of that kind of stuff. Tell us a little bit more about, you know, what surprised you about that, what things have had to stay human and. And how have people kind of responded to the changes they’ve had to make?
15:24
Taylor Bradley
The biggest surprise to me is a project that were at the London off site working on, and we’re calling it TalentBridge, internally is the name of the project. And we recognize that we, of course, use an enterprise HRIS system, but there’s much of what our talent flows do that is so unique to us that we need a system that can track the full life cycle and provide managers certain mechanisms that they can engage with it. So without getting too detailed on the concept, that’s the tldr, if you will. And we knew we needed this, and weren’t sure where to begin. I started to talk with my Turing Intelligence custom engineering team, and then one day, unexpectedly, I had our talent operations manager come to me. His name’s Arnaldo. And no background in coding, no background in engineering.
16:22
Taylor Bradley
And he said, hey, I want to let you know, I built a proof of concept for our platform. And he brought it up on his screen, and I was expecting maybe a slideshow or something along those lines. But what Arnaldo did is he connected with our infosec team. He was able to get a secure license to a tool called replit. And he built the platform with Gemini, actually in replit, and built the platform, or at least the skeleton of it himself, without writing one line of code. And that blew me away, is not only do you have someone that’s excited, that’s engaging, that’s pushing the boundaries of their skills, but they literally built a tool that were able to take to the engineering team and say, here’s exactly what we want. It’s already about 70% built.
17:16
Taylor Bradley
Now we need engineers to come in and actually scale it and bring it into our environment. And that was something that just jumped out to me of if that is where we are today with AI. I cannot fathom where we’ll be five years from now.
17:33
Matt Alder
Absolutely. That’s exactly what I’m going to ask.
17:35
Matt Alder
You to do now.
17:37
Matt Alder
So, I mean, what do you think the future might look like? I mean, how could things evolve with AI, with talent acquisition over the next few years? What kind of vision are you thinking of based on everything that you’ve seen?
17:50
Taylor Bradley
Yeah. So I will qualify my point of view right now with, pending any watershed moment in quantum computing or general intelligence with AI. Set that aside. If that happens, then. Unclear. Right. But at this stage, when I attended the CNBC CEO Summit, I’m on CNBC’s Workforce Executive Council and I have the privilege of attending the CEO Summit every year as a representative of the Workforce Council. And a year ago, CEOs were clamoring of what is this AI thing? Where do we even begin? This year’s summit was entirely different. It was, we’re starting to see individual use cases in different departments really start to gain traction. How do we scale it? I think next year’s is going to be, now we are seeing all sorts of different AI use cases in our organization.
19:00
Taylor Bradley
How do we start to allocate budget in the most efficient, effective way? How do we identify, hey, this use case is fantastic, but right now the amount of compute it’s going to take outweighs the cost. It would have to have a human do it. Right. And so I think we’re going to get to, at least in the next 12 months, these more meaty questions around budget allocation, capital allocation, which use cases need to be paused until maybe there’s some further advancement in AI or at least a more commoditization cost drive down with compute compared to where IT is today. The leaders that I speak with, the human resources leaders that I speak with are primarily bought into the fact that they need to use AI.
19:53
Taylor Bradley
I still think there’s a lot of work around breaking down this wall that artificial intelligence is an engineering or an IT initiative. It’s not. And I go back to the example I shared before with Talent Bridge. We built that, we productized that, we did work with our custom engineering team, but it was after that we had already built the skeleton and the foundation for it. And then the third macro thing I would say is I think right now with a. And I am not a lawyer, so this is not legal advice. I think as a spectator of what’s happening in the legal round is we are still in the Napster phase of the music industry. If we go back to the early 2000s where Napster was getting sued by everyone in the world for copyright infringement and all these different things.
20:49
Taylor Bradley
And then that actually evolved into music streaming and now you can’t find a CD or for those that don’t know what that is a compact disc if you wanted to. Right. And so I think that right now we are seeing organizations responding to AI with litigation. But I think we will get to a point where you have these major movie players out there that will start to figure out how do we actually make our own LLM? How do we monetize this instead of necessarily fighting the industry on it?
21:26
Matt Alder
Absolutely. Taylor, thank you so much for speaking to me.
21:30
Taylor Bradley
Thank you for having me.
21:32
Matt Alder
My thanks to Taylor. Don’t forget, if you haven’t already, you can benchmark your talent acquisition capability quickly and easily by completing the free Fit for the Future assessment. Just head over to Mataulder Me Podcast. It only takes a few minutes and you’ll receive valuable insights straight away. You can follow this podcast on Apple Podcasts on Spotify or wherever you listen to your podcasts. You can search through all the past episodes@recruitingfuture.com where 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.
