Diagnosing The Workplace: Not Just An HR Podcast

What Is The Impact Of AI On HR?

Roman 3 Season 2 Episode 6

Send us a Message!

In this episode, we examine the realities of artificial intelligence (AI) in our workplaces, specifically looking at its impact on Human Resources. We break down AI into practical terms with the intent to make its use and integration into our workplaces more practical and meaningful.

Our prescription for this episode: understand what is and is not AI, as there are a lot of misconceptions out there. Also start thinking of AI as a productivity tool, that is like adding another member to your team.

Statistics referenced:

49% of HR professionals say they leverage seven or more employment systems of record, inclusive of their HRIS, ATS, benefits administration, payroll, and time-tracking systems.

38% of HR professionals whose organizations have more than 1,000 employees report having ten or more employment systems in their tech stack.

The vast majority (84%) of HR executives at the VP level or higher (e.g., Head of HR, CHRO, etc.) believe their teams are using generative AI, yet only 34% of individual contributors report doing so.


- Finch. Early Look: Finch Survey Finds Many Opportunities to Improve the Employment Tech Stack. (2023)

About Our Hosts!
James is an experienced business coach with a specialization in HR management and talent attraction and retention. 

Coby is a skilled educator and has an extensive background in building workforce and organizational capacity. 

For a little more on our ideas and concepts, check out our Knowledge Suite or our YouTube Channel, Solutions Explained by Roman 3.

[ANNOUNCER]:

Breaking down everyday workplace issues and diagnosing the hidden sickness,  not just the obvious symptom, our hosts, James and Coby.

[COBY]:

Did we lose a patient?

[JAMES]:

No, that's just my lunch.

[COBY]:

Hey, thanks for joining us. I'm Coby. He's James. So let's get started with a question.  What is the impact of A I on human resources?

[JAMES]:

OK. Before we get started, I, we gotta make it clear. We are in no way claiming to be  AI experts, unlike probably about 25% of your LinkedIn connections, but we do have insights  into this topic and I think it is an important topic because we've got significant expertise  around employee performance and productivity and we believe that these AI tools can have a  significant impact in these areas. So I guess the question still is what's AI's impact? I  don't think it's going to be the massively disruptive, all encompassing replacement to HR  functions that I've seen spouted off in articles on LinkedIn or on Reddit. I think there's a few  things that we need to clarify right from the beginning and that's what it is, what it isn't,  and how we can use it. And I think the first thing I want people to understand is that  AI, as it's talked about today, it's not artificial intelligence. As we've grown up  thinking about what AI would be from movies and TV. There's no actual intelligence involved. The  tools that we have can do some really cool things around, you know, analysis, predictions, making  recommendations. But those are very different things from actual artificial intelligence.

[COBY]:

Yeah, I mean, I think it's kind of like the difference between what AI is today and what  we thought it would be by 2023, is kind of like I'm gonna date myself here, but it's kind of  like what hoverboards are today and what Back To The Future told us hoverboards were gonna be by

[JAMES]:

Yeah, well, as a kid in the nineties,  I was really looking forward to the 2020 hoverboards.

[COBY]:

Absolutely. So was I, so I do think that it is an important thing to realize that,  you know, that there's a lot of cool stuff that can come from AI, it is impressive.

[JAMES]:

I mean, let's, yeah, hey, I love some of the AI tools. I geek out on them. I use them  all the time. But I think people need to, we need to be clear in what we're saying. It's  not intelligence. Because intelligence implies that there is some learning some application to  that. And like, yeah, it is really and it really smart, really advanced programming. And let's not  diminish the accomplishments that people have had, who have created some of these algorithms  and the tools and technology that we have because they are phenomenal, but they're not artificial  intelligence as society has learned to think about AI. And I think that because like, again, there's,  there's fear around AI or some apprehension to, you know, like kind of fully embrace it.

[COBY]:

I mean, my concern is more like the buzzwordy side of it because like, I mean,  one thing that drove me crazy few years ago when, you know, like crypto became the... you know,  like half of your LinkedIn connections were crypto experts and NFTs and that kind of stuff. I mean,  like it's not, you know, I'm always afraid of AI in what, you know go down the road of it being  like, you know, you turn the blockers on when you see someone's, you know,  NFT crypt expertise or this kind of stuff like that because like you say, there's a lot that can  come with it. But it's important to not be fooled by the buzz around AI because sometimes we use,  you know, like automation and stuff like that is sometimes thought of or is presented as AI or like  even like speech to text software that doesn't have any kind of analysis component. It's just  straight speech to text. People think of it as AI but, you know, there's a lot of like, of software  and programs that are, that are pitched to HR professionals and business owners that is AI  powered or it's AI integrated. But really they've tacked on a AI component that's  really kind of like adding a digital clock radio to your toaster. Yeah, it's cool,  but it doesn't change how toast is made. So it is important to realize that there is this,  the buzz that it gets a bit misused or more common things kind of get lumped into it.

[JAMES]:

Well, I mean, it's, it's like any AI is the big boom right now, like with any big  transformative technology or otherwise, you're going to have the founders,  the people who understand the technology who have created it,  who have done some amazing things. And then you're going to have the opportunists who  present a good case and are very good at selling at convincing people of one thing and who kind of  steer the conversation or pull some legitimacy away from it. I'm, I don't think AI is going to  go the same road as like "AI expert" isn't going to end up with the same type of roll your eyes as,  you know, crypto expert or NFT expert. I don't think that's going to happen because,  well, NFTS were garbage anyways. Well, I mean, it was just dumb, right? It was, it was dumb.  But AI has like the tools that we are talking about when we say AI, you know, the analysis,  the prediction, the recommendations, these have value, they have legitimate value and  they can do some really cool things in our organizations. So expertise around that  should be celebrated. And I don't think it's going to have the same impact that those other ones did.  I think we are going to have to get a whole lot better at verifying expertise around AI.  Because right now, most of what I see is clever marketing, not actual AI.  Yeah, like you, you've said it before with, you know, kind of these tacked on features,  it seems like every tech company or the software company out there right now  is adding an automation feature and calling it AI and automation is not like automating a task  is great. That is a productivity tool. If you can create automations, you should. But again,  it's not AI and they're just slapping the label on in order to jack up prices and  to drive subscriptions because everyone wants to be at the forefront of the AI.

[COBY]:

So I think it is really important for people to just kind of take a bit of a step  back when you're looking for this kind of thing and realize that you kind of have to have a bit  of a grain of salt when you're looking at AI stuff because I wouldn't go as far to say that there's  AI snake oil out there. But I would go as far to say that you're right that AI, because there's  not been this universally defined, ironclad definition of what constitutes an AI advance or an  AI integration that were subjecting, We're being, we're being subjected, sorry, to some to a lot of  really clever marketing. So I just think that, you know, it's important to realize that, you know,  automation again, speech to text without, you know, any kind of analysis or the, you know,  slapping on an AI component to an existing software that doesn't really add anything to,  as far as its, its analysis component to it. Those are just  AI marketing and there's a lot of that out there.

[JAMES]:

So I think it's just more of a cautionary piece for people just to be on the lookout for  that because, and I think it's fair to say like I think your point of there is not currently  one standard definition of when we use the term AI, in regards to business, it means X,  Y, and Z and because that's not there, it's kind of swung the gates are wide open and anybody can  call what they're doing AI, I understand that. I just find it, I still, I find it very frustrating  and intentionally, if not intentionally misleading, then intentionally confusing.

[COBY]:

No. And I think that's a good point. So let's OK. So that's really  what it is. And hopefully we've put enough information out there that to,  for people to just think critically about what they're looking at when they see that AI sticker  attached to it. But what is it actually, well, at its heart, it's really just a productivity tool  and that's kind of the, the thing that you really have to think about it as it is a cool tool in  your toolbox and we can really get some great value of it. I mean, like, you know, like you say  James, you use it all the time and some that, you know, can really help us in our productivity side  of it to improve the, a more or improve how we use our time and improve how where we put our effort.  And I think that there's, and there's so much of a beneficial impact that can come from that.

[JAMES]:

Yeah, I mean, the way that I use it right now the, my primary AI tool that I use,  like probably many people is chatGPT I enjoy using it. It is a great  tool to help me to generate some initial thoughts. And for me that I work best when I have something  that I can deconstruct something that I can, you know, having a first draft or having  an outline or something that I can tear apart and then put my, you know,  build, rebuild from the ground up. So using something like chatGPT for my first draft or  for it to spark ideas or to you know, generate that something that I can destroy and rebuild.  I love it for that. And I'm on it all the time and it's really been for me, a really useful and  beneficial tool because it's saving me hours of time of having to frame, create my own outline,  pull it out of my head and get it down on paper and then deconstruct that I can have  it done for me in about 30 seconds just by asking the right types of questions.

[COBY]:

Well, yeah. And we use tools for like when we have our, our client meetings that we have,  you know, dictation tools that sit in on the meetings and generate kind of summaries and, and  action points and, and those kinds of things. And so like, yeah, so there's lots of little things  that we do that and it really is just about like, you know, getting something started or getting  something out there or getting something written down for us to build off of. And that's, and that  really is where again, like, you know, we're not a massive corporation that has, like, you know, that  needs it integrated as far as our talent analytics and as far as our, ATS and stuff like that, like,  you know, the we're not big enough that that is that, that we use it in those capacities, but we  do actually work, you know, in the realm that requires people to use that sort of stuff. But  at its end, what the way that so I talk, you know, I panel discussions and I do webinars  and I speak at conferences, you know, this kind of thing and when this kind of comes up,  the way that I try to frame it for people is just, is it is kind of a helpful way of kind of breaking  it down into kind of three buckets. I usually kind of say that we're talking about what AI does,  like what it is is that it performs really these three functions, it analyzes and predicts. So it  collects information, looks it over and does like a predictive analysis or says kind of, you know,  this is by looking at this year's data, this is what next year's data is gonna look like.  Or you know, those types of things are kind of predicting pieces that will come in the future  it analyzes and it creates, which is kind of like what chatGPT does it analyzes your prompts and  kind of the existing kind of databases that has access to and then generates or creates  something that will you know, give you a draft of something for you to work off of. And the  third thing is it analyzes and it recommends so look over all of the existing ideas and,  and data that you might have and like recommends it kind of like ranks them  and recommends the top one. So like a good example of that is when it scans resumes, scans,  all the resumes have been submitted into your, into your ATS or whatever. And then it recommends  the top candidates that fit the prompts that was put, that was put into it. And that's really like  when we say kind of like 85% of the AI that's out there that we want to use. It really does  do those three things, analyze and predict, analyze and create, analyze and recommend.

[JAMES]:

And what I like about that distinction is like you talk about recruitment and this is  one area that has seen AI tools really be promoted hard because companies having to sort through 100,  205, 100s applications for every single position, it's just the man hours involved  are ridiculous. So being able to analyze and make recommendations based on the parameters  that you put in is a good thing. It's an automation tool that will help to save time.  My concern is always in how much people rely on the tool to do the job for them, right?  And I don't know if that's a not understanding the usage of the tool or how it's actually used or  if it's more of a, well, I just, I have 47 other things on my plate. So I'm gonna let this program  take care of what it's supposed to take care of and then just I'll use whatever it recommends,  right? So there could be a bit of that because one thing that I found really interesting when  we're talking about like HR and tech and AI is that just the sheer number of  tech programs that HR professionals have to use on a regular basis, like there was a, there's a new  study coming out October of this year. Well, actually, it might be out by the time this  goes live, we'll put a link to the to the preview in the show notes, but they found that 49% of HR  professionals say that they leverage seven or more employment systems  already. And even in larger organizations, so organizations that have more than 1000 employees,  38% of HR professionals report having 10 or more employment systems in their tech stack.  That's 10 different systems that they have to bounce between all of them, promoting different  AI systems, different AI tools, different features that they have to put in the information like it's  rightfully and understandably overwhelming the  number of different tools and systems that HR professionals are subjected to.

[COBY]:

So one of the things to consider too is that one of the big barriers to AI having this  massive, this disruptive impact on HR is gonna be the fact that we have all these programs that are  not talking to each other and the databases are so segmented that there's never gonna that until we  can actually, unless they like it really all speak the same language that all the data is fragmented  and this database doesn't have access to that database. So, I mean, you know, one of the,  it's kind of, it's funny the variety of a lot of these different, you know, like these 10 potential  different programs that we use around our HR is and ATS and all that other kind of  stuff like that is that if they're not talking, the inputs into the  AI engine and the AI databases is gonna be what's holding back the outputs.

[JAMES]:

Well, absolutely. I mean, AI is incredibly sophisticated technology. Like  we've already said that before, you can really make use of AI tools. You need to have really  good automations in place. It requires in order to have to be AI savvy, you need to be automated,  automation savvy. But HR is not even automation savvy at this point because they by and large,  we're not data savvy. Our data is not consistent, it's not clean, it's not like structured  consistently in a way that will allow for all of these other things. Now, there are absolutely some  players in the market who get this and understand it and are doing amazing things with it. But by  and large, our data in HR is not clean enough is not consistent enough to really allow us to  leverage what AI can and should be. So we're still using a lot, we're using a piecemeal  and we're not using it really to its potential. And I don't really see it getting to that point  before we're able to actually clean up our data and make it usable in these advanced systems.

[COBY]:

Well, I think this is the thing too is that a lot of what the executives want AI to do  to improve, you know, the again productivity and performance and leverage this incredible  technology for them to get what they want, they're gonna need to have let everything  be a unified database, right? For it to try to achieve the stuff that we really wanted to be  a game changer for our organizations. Then we have to convince, then we have to migrate from 10 piece  meal pieces to either one or two or to enough that can, that will talk effectively so we can  have this rich point of data where we can actually leverage it to what, what we say it's gonna be.  And I think that potentially we'll definitely get to a point where we could get to where we want  to be from a technology perspective first. But the organizations, well, you know, not be able  to get there themselves because they're going to have all this patchwork data everywhere.

[JAMES]:

So I'm gonna be replacing HR functions or HR people in the near or foreseeable future like,  I mean, yeah, sure, someday who knows where technology will go? But it's just,  it's not clean enough. It's, and I mean, if you're an HR professional listening to this,  you know how bloody important it is to verify data, right? If a single the the data that you  are collecting that you are storing is sensitive information and to completely pass that over to  an AI system without checks and balances is, well, it would be opening up the organization  to liabilities which is part of our job to mitigate those liabilities, right? Like it is  counter to the intent and purpose of HR in as a function of protecting the business. And

[COBY]:

I think the other thing too is, is also kind to keep in mind there's,  there's a bit of a disconnect between how we can use AI functionally and practically today and how  we do actually use it, right? Like, there is this, there is this, this disconnect where, you know,  some people I think like a lot of people think that their teams or even their competition are  doing so much more with AI than they actually are. So, so there's a bit of that, but I really  kind of think we need to kind of all just take a, take a step back and realize that there isn't  a massive discrepancies between AI usage, you know, between people or it's really  probably not being leveraged in a way that people think of this.

[JAMES]:

Well, and what's really interesting is that same,  that same study that I referenced earlier had a stat around senior leaderships perspective. And I  love the idea of, you know, we think that people are using it way more, whether it's competitors  or colleagues, this idea of "AI Paranoia" and the they found that the vast majority, 84% of  HR executives at the VP level or higher believe that their teams are using generative AI, but only  34% of individual contributors reported doing. So that's a massive discrepancy between what senior  leaders believing or fearing that their teams are relying on AI technology versus the, I mean, 34%.  Acknowledging that. Yeah. Well, while I do use these generative tools.

[COBY]:

Yeah. And, and part of it too, it might be that the general tools that they're  using is I'm using chatGTP to help me get, you know, the first draft done or I'm using  some of the communication editing stuff to kind of help make my, the tone of my email a little bit  more diplomatic or a little bit more confident, you know, I mean, like there's so, so it could  just be as simple as that. But, but the paranoia at the, at the c suite level could almost be like  the same as like an English professor thinking that all of their, all, all of their students  essays are all AI generated and no one's actually doing any of the work when reality is like,  you know, it's, it, there, there's not this massive, you know, plagiarism piece. So, I mean,  so it is really important to realize too that you, right. I think there certainly is AI Paranoia  and I think that we almost need to just accept that our fears are not necessarily the reality.

[JAMES]:

So they also need to put checks and balances in place to  make sure that our fears don't become the reality.

[COBY]:

Excellent point. So I think that from there we should go, we should really move into.  Ok. So we talked about what it isn't and we talked about what it is, but how can we use  it? And, and I think that if to really kind of reiterate what I said before, when, what it is we  need to use it as a, as a tool, which is what it is. It's a virtual assistant. It's not a decision  maker that we need to be leaning on or relying on or it's not the scary robot that's gonna take over  jobs. It is a tool that can probably function at the level of a low experienced virtual assistant.

[JAMES]:

...or intern. I mean in many respects,  AI tools are your intern. They don't do anything unless you specifically tell them to,  you need to give them very, very clear instructions on what you want  and you need to verify the work that they produce.

[COBY]:

Yeah. you're right. That's,  that's a really good way to put it that it's like having a really smart intern.

[JAMES]:

A very smart, not socially adept intern.

[COBY]:

Yes. I mean I suppose it kind of takes you back to, like the, you know, workplace comedies  where we have these lowly interns don't even have their own name that we're always like yelling,  "hey intern" or intern do this or, you know, giving this unpaid college student like,  you know, the stack of paper and say, you know, review all this and give me a summary by...

[JAMES]:

..by 8 a.m. tomorrow morning. Yeah, so I mean that  we are not advocating the abuse of internships.

[COBY]:

Actually, I think that maybe what we could say is that if AI is your intern,  then maybe we could actually give our interns a real experience and actually have them,  you know, developing the skills that they need instead of giving them the grunt work that we  don't want to actually do, we can give that to AI because AI is not going to complain.

[JAMES]:

Well, I mean, and, but I mean, look at the example that we talked about,  we talked about recruitment earlier, right? Would you let your intern make your decisions for you or  would you tell your intern here are the criteria that I want you to assess these resumes for,  organize the in what you believe are the, you know, top 10, top 50 top whatever and then  verify that. I mean, yeah, make, make AI do the repetitive, the grunt work but verify it. Right.

[COBY]:

Yeah. Absolutely. Well, I mean, and I think there's actually a lot of value in  thinking of AI like your intern because you know we even we go back to the analyze and predict,  analyze and create, analyze and recommend, you know, like if you're looking at  like, you know, analyze and I don't know, pre or sorry and, and recommend you're like  hey intern scan the legislation for references to return to work accommodation timelines. So we can  and recommend the citations, you know what I mean that you still have to verify that like you would  with an intern, but that's a good, you know, but that's a good use of it, right? Or yeah,  or as far as create, you know, hey intern what is the, what is the Excel formula for calculations  that ignore blank cells, right? Like, I mean, I almost like have it, you know, give you that  kind of, you know, generate or create that, that walk through for me so I can do it because I don't  know how to do that, that one thing myself and I don't want to go looking it up, right?

[JAMES]:

Two things. One, the Excel thing has saved my butt so many times. I absolutely love  it. Give me an Excel formula that does this, this, this and this and it just. the other thing is the  reason I laughed when you talked about legislative is I remember the story from a couple years ago  and I just, I pray, I hope that it is true. I haven't been able to actually verify it about a,  lawyer who was preparing for a defense and used Chat GPT to write his opening statement,

[COBY]:

Oh, a couple months ago.

[JAMES]:

Yeah, with the citations and everything and, you know, citing,  precedent and all these things and presented it in court as is, and cite it,  cited past cases that don't exist. Right. Check your interns work. Absolutely. But I mean, and again, even somewhere around the ideas, you know, of like scanning,  you know, new employee information to kind of put into your HR is like, "Hey Intern, enter this new  hires data into our system", right? Like, I mean, there really are you know, a lot of the uses, of  what AI can do is kind of like you say the grunt work that we pass off to an intern. So I really  do think that there's a lot of value in thinking of it from that way, like if you were looking for  an intern who can analyze and predict, analyze and create and analyze and recommend,  then AI is that intern for you. But at the same time, they are just an intern, like the same  reasons interns are not running our HRbdepartments are the same reason why we don't want AI running  our departments and making decisions for us. So I think that it's a very apt comparison and I think  that it actually might be a pretty... I love the intern analogy.

[COBY]:

I, yeah, well, actually, even if you want to go one step further, what if we are like,  instead of calling it "AI", what if we call it? "Hey I" for, "Hey Intern", huh?

[JAMES]:

No.

[COBY]:

Or this is a way of thinking like...

[JAMES]:

just sometimes man.

[COBY]:

OK, maybe we'll go that far, but just maybe

[JAMES]:

Maybe that one. So let's comedy retract.  We'll stick with the internship analogy and everybody can just ignore the terrible joke.

[COBY]:

But I do think that, you know, actually I'm gonna double down on that.

[JAMES]:

Of course you are.

[COBY]:

Well, because my thought was like, you know, we said before that AI is not the  artificial intelligence that we thought it was from, you know, years ago. But what we're actually  experiencing right now is more like "Hey I" where we're actually asking or actually demand or giving  prompts to our intern to do this grunt work for us, which is not the, you know, movie AI, you  know, that we expected it to be. So it actually I do think there's actually some merit in "Hey I".

[JAMES]:

Yeah, there's, there is some merit in it. Well, but, but we'll,  we'll move on. I don't think it's going to be catching on quite as much as you hope.

[COBY]:

Rats, anyway. But I do think that this idea of how can we, again, how can we use,  AI in its, in the impact of HR I really do think that, you know, it, you know, again, we won't  necessarily have the ability to migrate all this stuff at this point. You know. because like we  have all, we've invested in all these different programs of software but for task use is probably  the best way for us to think about it, right? When we're trying to again, do things around, analyze  and predict, analyze and create, analyze and recommend having that like, you know, almost like  intern support to do some of the stuff that will be more time consuming and it will be more tedious  to kind of speed up that process is the component that we should be looking for. So, so that's  really if we're looking to implement something or we want AI solutions for our current problems  that may be the way to go is to look for. Where do we need that intern level  authority and support to analyze and predict, analyze and create, analyze and recommend.

[JAMES]:

Yeah, I mean, AI is not going away. these tools are, it's, it's not something that  is a flash in the pan. This is, there will be some sticking power to the tools that  are being created now. We need to accept that and we need to figure out how to best use it.  And I like the way that you frame it around the three distinctions. I like our analogy of  you know, treat AI as your very intelligent intern who needs very clear directions and verify their  work. I think if you keep those two main, if there's two big takeaways from this conversation,  I think it's the analyze and predict, analyze and create, analyze and recommend. And it's the  treat AI as an intern and those two pieces will help you to really filter through the  just the sheer number of AI tools that are being promoted to you probably on a daily basis.

[COBY]:

And I would also say, you know, and to realize that a lot of  the impacts that AI is gonna have on HR and our businesses kind of as a whole is gonna be  probably held back by how we do things currently. So again, like the quality of the data that our  software has access to will be probably one of the biggest hurdles to the kind of advances and the  improvements in the productivity and performance improvements that A can provide, you know. So,  so it is something that we need to really keep in mind too that adding a bunch of new bells and  whistles on to stuff may actually complicate the problem. So we wanna make sure that,  you know, if we're trying to have a more strategic plan of how we're going to use  and leverage AI's ability that we're realizing that the messier we are with our data and the  "set it in a forget it" approach that we may be thinking we can get from AI where it where it  will be able to make decisions for us. Those are things that we have to just kind of realize are  gonna be major barriers because we never, we, we never want AI to be making decisions. But we also,  if you want to leverage it, we have to make sure that there's the quality of clean data  that we'll be able to pull from because again, the outputs are only gonna be as good as the inputs.

[JAMES]:

Yeah. Well, I mean you're absolutely right because you're, we're talking about  just creating big, huge massive databases of information for these tools to pull from. And if  your, the information that you get out, you're right is only as good as the information that  is in there. So if your information is not well organized, it's not clean, it's not consistent,  you know, your systems don't talk to each other. You are going to struggle with the more  productive, more advanced features that AI can provide.

[COBY]:

Yeah. So when teams try and use it in its current piecemeal form.  you know, the limitations are gonna be around the analyze component of analyze and predict and  create and recommend, right? Because I mean, if you can't analyze things well, the prediction is  not gonna be as strong or the creation or the recommendation. But if the teams like I say,  are trying to use it to improve the performance of productivity, then realizing that it's about  leveraging support, like an intern, to help to take away some of the more time consuming TDAS  components to kind of get you the way there like the way that you use it to kind of say,  you know, I have a hard time just putting pen to paper to get the first ideas down for when  I write something. But I'm really good at once. I have something to break it apart and figure  and make it, make it the way I want to that, you know, time consuming piece of getting that first  draft out is how you use it to improve your productivity. But that component could be,  but that idea could be translated into our teams and into our work is that if we're again getting  it to a point where we're using these tools, whether they're gonna be piecemeal for a long  time that allow us to kind of get that over the hump of a first time consuming component,  whether it's reviewing all these resumes or whether it's reviewing all these benefits or  whatever it is like, that gives us something to work off of. Instead of starting from scratch,  we're kind of starting halfway and that is gonna be the impact it is gonna have on HR is that  it can, it can help us get to halfway a lot faster than the old way without that intern support.

[JAMES]:

Yeah. No, that's, that's a good, that's a good point.

[COBY]:

Yeah. All right. So I think, I think that pretty much summarizes the conversation.

[JAMES]:

I hope we didn't confuse people too much and I hope we didn't lose anybody with your jokes.

[COBY]:

Yeah. So if that happens, I really apologize. But yeah, so I do a bit of a summary  but I really hope that we were able to provide some type of tangible clarity for folks because  I do think that again, this is a really important thing that we can't ignore, but there's so much,  not misinformation, but confusing information out there that I think providing a bit of a really  tangible kind of clear, practical way of looking at this complex tool should really help people be  able to kind of figure out how, what does this mean for me and how can I use this for me? And  I really hope that that's something that people are able to take away from this conversation.  So, what is the impact of AI on HR? Well, first it's important to be clear about what it isn't,  it is not the advanced, you know, robots that we thought was gonna come from TV and movies and that  it's also not automation or simple speech to text. It is really about the concept of  analyzing information for our use because what it is, it's a productivity tool. It's about helping  us analyze and predict information to kind of see what's coming in the future, analyze and create to  generate something that we can use to kind of get us to halfway to where we need to be or analyze  and recommend from a large body of content to figure out what is the best place for us to start.  There is some paranoia around how AI is being used by other people. We have executives concerned that  their staff are relying too heavily on it. And we also are afraid that our competition is using it  far better than we are. So there is this kind of pressure that we put on ourselves around AI  and that's also kind of impacting kind of HR and the people involved to almost like, you know, do  better than the company next door. But the really important thing is how can we use it? Well, again  AI is a tool. It's like a virtual system. It's not a decision maker. It is like having a really smart  intern that you can kind of yell, "Hey Intern, put this new information for a new employee into our  system". It really is a boat giving some practical work off our plate onto somebody onto something  else that can actually, you know, save us some time and save us some some of our productivity  and though James hates it, I think there's validity in us thinking about AI more like,  "Hey I" and think of it as what would I yell? "He Intern" for and those are probably  the things that would be functional. I don't think it's able to get coffee at this point.

[JAMES]:

Oooh, AI delivering coffee to me every morning.

[COBY]:

But ideally and hopefully this, that comparison might actually allow for interns to  have a more effective and experience when they're, when they're trying to get the college credit or  whatever. But ultimately, we wanna make sure that we are leveraging this great tool to improve our  productivity, to improve our performance and have a significant impact on our workplaces. But until  we are able to kind of get our data clean and have the stuff that our AI is is analyzing to be robust  enough, we're gonna be one of the biggest factors that slows down its integration in our workplaces  because AI needs to have a strong consistent database in order to provide that prediction,  that creation and the recommendations. All right. Yeah. Anything else that James?

[JAMES]:

No, I'm good.

[COBY]:

All right. So that about does it for us. For a full archive  of our podcast and access to the video version hosted on our YouTube channel,  visit www.roman3.ca/podcast. Thanks for joining us.

[ANNOUNCER]:

For more information on topics like these, don't forget to  visit us at www.roman3.ca. Side effects of this podcast may include improved retention,  high productivity, increased market share, employees breaking out in spontaneous dance,  dry mouth, aversion to the sound of James's voice, desire to find a better podcast...

People on this episode

Podcasts we love

Check out these other fine podcasts recommended by us, not an algorithm.