Transcription
Emma O’Connor: 00:08
Hello and welcome to another episode of DC Talks, a podcast series from law firm Doyle Clayton, where we take a dive into an HR, a leadership or a legal hot topic and for today’s podcast, we are discussing the topical issue of AI in the workplace. Just a reminder that this is a podcast, we are not giving you any legal advice or assistance, but if you would like to find out more about how we can help you, the information will be in the show notes. So, for this podcast, I am joined by Rosanna Proctor, who is a senior associate here in our employment team. Hello, Rosanna.
Rosanna Proctor: 00:54
Hello.
Emma O’Connor: 00:55
And also I am joined by Dave Jackson, who is Head of Technology again at Doyle Clayton.
David Jackson: 01:01
Hello.
Emma O’Connor: 01:02
Dave, welcome.
David Jackson: 01:03
Thank you.
Emma O’Connor: 01:04
So, today we are discussing ‘AI in the Workplace: Tips & Traps for HR’ and I think what would be a really good place to start is actually thinking about AI in its more general sense and Dave, if I can ask you first. What is AI and why is using AI different to maybe using a jungle internet search?
David Jackson: 01:32
Yes, AI is not new. AI has been around for decades but what is new, is how accessible it’s been made to us all and how at everybody’s fingertips it now is and that obviously came about a few years ago when ChatGPT really made headlines and everybody suddenly jumped on the bandwagon and was able to start using it. And it’s funny really because we often think of the internet and internet searches, if you say you’re going to search the internet, most people will say, I’m going to Google it and we’re pretty much doing that with ChatGPT. It’s kind of synonymous with sort of searching AI and using that as a tool. But AI is generally like a broad umbrella that covers systems where you can analyse data so that might be CV screening or HR analytics.
David Jackson: 02:22
You might want to generate some content, so whether that be text, image, audio but you can also make predictions or recommendations, you know, with data that you’ll be using in AI.
Emma O’Connor: 02:32
So as we know, this is an evolving topic and also evolving technology and the aim of our conversation today is really to start HR teams, businesses, thinking about maybe some of the advantages of using AI at work, but also some of the tips and obviously some of the HR traps as well and I suppose if businesses are at the kind of the start of their AI journey and many might be thinking how they can use it, how it might benefit their businesses, I think those employers who are saying their staff aren’t using it, probably I would imagine most employees are using it, even if they haven’t been sanctioned to use it.
Emma O’Connor: 03:17
So we think about uses within the workplace and if we think particularly around HR uses of AI, can you give some examples, Dave, of how HR teams might be using AI?
David Jackson: 03:29
Yeah, absolutely. Yes, I suppose the classic and easy one is probably drafting job descriptions so obviously to help you with putting those together. CV screening and ranking so you put a job description out there, a job spec, and then you’re going to get inundated with hundreds of CVs. You’re going to need a quick way to root out the good candidates or potential good candidates from that. They might be using it for interview question generation and then probably some analysis of data. So you could analyse surveys and maybe do some internal surveys and you want to do some analysis of the input that you kind of get back from that and maybe even things like spotting absence and turnover trends as well, I’m not sure your employees. So they’re probably just a couple of simple use cases that I think HR teams would be using it for.
Emma O’Connor: 04:16
Absolutely, and I think certainly the spotting absence trends is obviously a real HR hot topic at the moment, particularly when we think about the changes that are coming in April of this year around changes to SSP. As we know, with all benefits comes risks, doesn’t it?I think it would be helpful to discuss some of those risk areas Again, not just from an employment law perspective, but also maybe sort of in its wider context as well, bringing you new technology, what that might mean in the workplace. So, what are you finding in terms of questions you’re getting from clients and possible risk areas?
Rosanna Proctor: 04:55
So I think certainly there’s, as Dave mentioned, there’s lots of ways that the tools can be used to make work more efficient for HR teams but with that, as you say, does come risk and I think certainly one area is the risk of bias and potential discrimination. So to use an example, thinking about CV screening, that could potentially be seen as a tool that creates a more diverse pool talent, less discrimination, but it really depends on what information is being fed to it. So, it must be used in the right way. The big concern is that if AI recruiting tools are using company data that has passed data that is potentially discriminatory or favoured one group of individuals, then that data will then be used going forward so it’s not without risk and it could potentially mean that there is discrimination going forward for that CV screening.
Emma O’Connor: 06:00
I think you’re absolutely right. The bias point is a very interesting one, isn’t it? Because a lot of businesses will think it’s okay reusing a computer or a software package that will reduce incidents of perhaps bias. But actually all you’re doing is you’re just sort of swapping potentially one of the devices for another and as you said, it’s output is only as good as the input with these systems.
Rosanna Proctor: 06:27
Yes, and I think on that, I think there’s definitely then something to be said if the HR teams who are using it aren’t aware of how the tools are working in the background. There’s a potential bigger risk that if asked to justify decisions, they actually don’t have or potentially don’t have those justifications there because they don’t have the knowledge of how the AI tools are working. Som there’s that added layer as well of potential risks and defend any kind of discrimination claims.
Emma O’Connor: 07:00
Yeah, absolutely. And I suppose the other point is you need to tell candidates and employees you will be using AI as part of recruitment or employment decision-making processes. So I guess that’s going to mean not just looking at policies around e-mail, internet and AI, but also looking at data and data protection policies and obviously privacy notices as well.
Rosanna Proctor: 07:22
Yes, exactly, and that’s a key thing that lots of HR teams are contacting us about at the moment in terms of updating policies. Data protection policies, as you say, but also IT and communications policies as well, just so that they are all up to date and crucially, everyone is aware of the policies, what they say, and how they can use it.
Emma O’Connor: 07:48
Absolutely. What about hallucinations? I hear hallucinations a lot, what’s that and what does it mean? David, do you want to sort of explain what hallucinations are and then maybe Rosanna tell us about some of the legal risks there.
David Jackson: 08:01
When we think of ChatGPT or Copilot and we input a prompt into them and we ask them a question, we’re expecting something back out of them. They give you an answer with such confidence that you would believe whatever it outputs. Hallucinations will normally be where the AI platform that you’re using has generally made something up. That might be because it’s using some sort of data, or reference to data, where you’ve asked it a question and it will make up maybe a, let’s say if it’s legal related, it might be like some sort of case law or something, it’s going to make up something that it thinks has happened when it hasn’t and you run the risk then, of taking that output from whatever AI platform it is that you’re using and running with that, believing it and maybe putting that into your own documentation, sending that on to someone in an e-mail, thinking that it’s true. There are things you can do to limit hallucinations when it comes to AI and the platforms that you use and that comes down to really good prompts. Being able to prompt properly is a real key training point and that’s something that, firms should really consider when it comes to the use of AI. You need to teach people how to use it properly so that the outputs that you get are more reflective of true facts and real life.
Emma O’Connor: 09:31
So when it’s needs to please, you know, AI’s need to please and it’s been programmed to part with an answer, that’s what it can come up with an answer, whether it’s right or wrong. Rosanna, I can see some really, really risks with that.
Rosanna Proctor: 09:46
Yeah, exactly that and as Dave said, I think the concern is the certainty of the output with what it’s the information it’s providing. So individuals, whether that’s HR teams or just the wider workforce, can get sucked into thinking that’s the surefire answer. So as you say, it means that it’s got this real potential risk that people could be essentially giving advice that’s not legally compliant or in accordance with any regulations, which will have massive impacts for the business and leading into potential claims. Just from my experience, interestingly, I find one sort of topic that seems to come up in practice that I’ve experienced in the last few months is a lot of employees seem to think they have sort of constructive dismissal claims and constructive unfair dismissal claims without the required service, which I seem to think is coming through sort of AI tools and that causes huge issues because employees are coming forward with those types of claims or HR teams can be thinking that those types of claims are potentially risky when in fact they’re not.
Emma O’Connor: 11:02
I think you’re absolutely right and we can talk a little bit later about the impact of AI on complaints and grievances as well, because I think that is huge and I suppose one of the other issues is around disciplinary and actually the employers citing AI misuse in disciplinary policies or in conduct planes. What’s your experience of that?
Rosanna Proctor: 11:26
This is another area that I’m seeing in practice in terms of organisations wanting to discipline for people misusing AI. However, there’s a real risk area there, if they haven’t got their policies in place and they haven’t communicated properly what the right use is. It’s so accessible now that people have it at their fingertips. So, to be suggesting that people won’t have access or shouldn’t use it, I think is a difficult line to take and what employers need to do is make sure that they have got the relevant communication and transparency out there as to how people can use it, so that if people are misusing it, and that’s a genuine conduct concern, they as an organisation have the ability to back up that actually there’s an employee here there that was well aware of what they can and can’t do but actually went against that and that can be a legitimate disciplinary concern.
Emma O’Connor: 12:34
I think you’re right. I think there’s two issues, isn’t there? And there always has been with e-mail, internet use. This is just another extension of, you know, communication policies that employers have had for a number of years now. And you’re absolutely right, the speed and accessibility of information at your fingertips, whether that’s right or wrong, is another matter. I suppose employers who think that their employees aren’t using AI, whether that’s as part of their job or using employer company systems for personal use of AI, if employers think that their employees are not doing that, I think it’s very much, you know, wake up, there’s a bit of naivety there. So, I think there are two issues. One, is ‘what’s our policy going to be on use within the workplace as part of your job and what are the considerations there? But also, as I said, as we’ve done with e-mail, internet use, social media use as well, what are the expectations around personal use? And I suppose some employers will have a blanket ban. Some employers may take the path of, well, okay, we allow some personal use, but it must be incidental It’s using our systems as a privilege. It’s not a right. It’s subject to monitoring and revocation and must comply with our policies as well. So I think there are lots of questions really that employees need to be asking themselves and asking themselves now about how they can use it and absolutely, those tips around having clearly communicated policies and that will depend, as you’ve said, on whether they are a regulated industry, say finance, banking, whether it’s legal as well, but also what the appropriate use is within the workplace and as we said, this is ever changing, ever evolving and I just wonder, Dave, from your perspective and obviously you are advising us as an organisation about our use and how we use it, as a regulated firm, but I’m just wondering, is there any tips that you have for businesses about how they can maybe future proof their use of AI whether that’s personal, incidental, or whether that’s professional use.
David Jackson: 14:48
AI is obviously the cool kid on the block at the moment, isn’t it, really? And everybody kind of wants to use it in some way or some fashion and I think if you’re an organization that wants to limit its use and say ‘we’re not going to use it and employees mustn’t touch it or do anything with it’. The problem with that is that you run the risk of driving it then underground and driving its usage underground. So, you’ve certainly then got to have systems and procedures in place whereby you can monitor, usage if that usage is kind of being driven underground. I think any business really should be looking at it in a more positive light, seeing where areas of the business it can be used on so that you can obviously demonstrate that to your employees. These are the things where we see some wins and where we see its use cases. The important thing is also to remember is that AI is not right for every situation and that’s again, got to be communicated, I think, to your employees as well. Just because we can use AI for something doesn’t mean it’s going to necessarily be the right tool for us, especially when it comes to HR as well, because I think HR is very much a human element to that and so we shouldn’t lose sight of that but I think certainly businesses, for future-proofing purposes, they really do need to start thinking about areas of the business that could benefit from it and really sort of engaging with employees to show that they’re on the front foot and they’re going to lead the way on the platforms because to be honest with you, it’s only an amount of time for employees, you obviously want to keep your employees happy and they want to work for companies that are hopefully forward-thinking with what they’re doing and if you’re not doing that, then you’re standing still.
Emma O’Connor: 16:28
Absolutely. Look, as we know, you must put your complaint, your grievance in writing. In the good old days, that was a letter. Then we had, that would be an e-mail. But now we are finding that employees are using AI, whether to raise a grievance or a complaint or even, a claim, rightly or wrongly, citing the law. Where has AI really had an impact, Rosanna? In terms of employees being able to raise concerns and are there any tips you’ve got for employers if they do get a 52 page AI generated grievances?
Rosanna Proctor: 17:04
Yes, which unfortunately is not uncommon in practice. But yes, I mean, seeing a lot of it, I think what it has been able to do is mean that employees have that instant access without having to pay legal fees to draft a document setting out every concern or claim they have under the sun. Setting out all sorts of legal claims which, as I mentioned earlier, may not necessarily be correct, but they’ll have a long, lengthy and tidy document to be able to submit. So, of course, that’s going to pile up the issues for HR to deal with, having to kind of unpack those grievances and concerns. I think the more that AI tools are used by employees, the easier it is becoming to potentially pick up when it is being used, particularly spotting things like hallucinations, referring to case law. Certainly in practice, we’ve seen claims come in that are referring to sort of case law that is just not correct. So unfortunately, we can’t get rid of them using those AI tools to submit those, but we can spot the trends and where potentially it takes more work to go through and really identify what those concerns are and what actually is relevant and accurate.
Emma O’Connor: 18:32
So it hasn’t changed in substance be approached to a grievance?
Rosanna Proctor: 18:39
No, I think you still need to go through your grievance procedures as you ordinarily would. Of course, if there’s some element that they have misused AI where they shouldn’t be within the workplace, then that’s a sort of separate issue but just taking forward a straightforward grievance complaint that’s been put in or a legal claim that’s been put in, you would still deal with that in the same way, but just being aware that potentially more employees may be submitting them because it’s easier for them to do so.
Emma O’Connor: 19:13
Rosanna, where you’ve got a business who’s maybe embarking on that AI journey, an HR team that’s maybe thinking about using AI in the ways that David’s mentioned, What tips would you give them?
Rosanna Proctor: 19:27
I think, as Dave mentioned, not all AI tools and not all businesses will need to use those AI tools going forward so it’s really about thinking as a business, what is useful to us? Where will it help us? Cause us to be able to streamline work and be more efficient with our work? How do we want to use that? identify that. Then separately, as you both touched upon, the separate issue of how the organisation wants employees to use AI personally and consider that. Can they use it? What can they use it for? What are the issues involved in that? And then as an organisation, being aware of the risks, I know we’ve briefly mentioned some of them today, things such as bias and misusing confidential information, potentially the hallucinations that can spring up. Being aware of all those risks and how we want to use that, doing due diligence into the ITILs that you want to use and then as a business, once you’ve thought about all of that, that’s when you can draw up your own policies and consider how you want to use that and communicate that with all your employees going forward.
Emma O’Connor: 20:48
Thank you very much. Thank you, Rosanna. Brilliant tips there.
Rosanna Proctor: 20:50
Thank you.
Emma O’Connor: 20:51
Thank you, David.
Emma O’Connor: 20:53
Thank you so much. Well, just to sum up, I think really, you know, there are positives, of course, to using AI, even that’s within HR with business in general. But of course, you know, with all those positives, of course, we have to think about some of the challenges as well. We’ve seen IT and systems evolve, haven’t we over time? And this is maybe a natural progression as well, but we do need to think about how we use AI. We need to think about having, as data sets, secure tools. We need to think about personal data. We need to think about client data. So again, going back to basics. the data protection and privacy notices. Again, when you think about regulatory issues, as Rosanna has mentioned, you know, policies clearly thought out, clearly communicated, and also making sure that employees are aware and they’re trained to how they should be using it. There are obvious benefits, of course, but with all things, remember not to take the human amount of human resources. For advice on policies, processes, and what things AI and employment law related. Again, please to get in touch until I get a chance to speak to you soon. Thank you very much for joining us today.