We now know that AI is already introducing major performance improvements in many companies. One trial this month (Jan 8 2024) conducted at the Boston Consulting Group by Professor Mollick at the University of Pennsylvania found that those in the group using ChatGPT4 completed tasks 25% more quickly and produced 40% higher quality results than those who did not.

Get up to speed on everything with our webinar, from setting clear business objectives to expanding AI into every relevant HR area and business role, and much more in between.

Digital transformation expert Kieran Gilmurray shares the ten steps you need to take to implement AI successfully in your HR department and beyond. Introduced and facilitated by Barry Phillips, Chairman of Legal Island

Kieran Gilmurray is the author of the book “The A-Z of Organizational Digital Transformation.” He is regularly ranked as one of the top global experts Automation, Data Analytics, Brand Influencers and Business Technology Innovators. He is a senior executive with 28+ years of experience leading digital transformations for businesses and has been a director, board member, research fellow and advisor to multiple international companies. And has won many national, international and global awards.

View/Download the SLIDES HERE:Using AI in HR Ten Simple Ways to Get You Started FINAL SLIDES.pdf and his helpful DEMO VIDEO.

Recording:

Transcript:

Barry:  Good morning to everyone. I hope you are well and have not been too badly affected by the storms overnight. My name is Barry Phillips, and I'm chairman of Legal-Island. And it's my huge pleasure to welcome you to this event, which is entitled "Using AI in HR: 10 Simple Ways to Get You Started". Can I say thank you immediately to our sponsor for today, which is MCS Recruitment.

I'm delighted to say our guest speaker today is Kieran Gilmurray. Kieran, who is based here in the island of Ireland, is an author on digital transformation and AI expert. He's worked in the industry for the past 30 years and for some well-known local firms, including Hughes Insurance, and international firms such as Pearson Education, Liberty Insurance, and Rapid7. He's also written or worked with IBM, Deloitte, Invest NI, and lots more big organisations.

Kieran, good morning to you.

Kieran:  Good morning, and welcome to everyone today. Let's have a bit of fun learning a little bit more about AI. Thanks for the invitation, by the way.

Barry:  Not at all, Kieran. It's always good to be working with you.

Kieran, right at the beginning, can we perhaps go back to the basics of AI? And let me just start with a few questions in that space. So can I ask you this? What is AI and how does it differ to generative AI?

Kieran:  Oh, gosh. I say that seems like the simplest question in the world and the biggest question in the world. But consider AI, or artificial intelligence, as a mathematical model that allows you to decide or predict something or produce some insights. Let me try and illustrate that with an answer.

Let me use Hughes Insurance, or an insurance example. Those companies have tremendous volumes of data, customer data, purchased data, renewal data, claims data, whatever else. If you use AI algorithms or mathematical formulas, whatever you want to call them, then I can look at that data, analyse it, and do very simple stuff that says, "Okay, how many customers have we got?" That's just descriptive analytics.

You could move a step further to diagnostic analytics and say, "How many customers left us in the last year and why did they leave us?" So we still haven't touched AI yet, by the way. I'm just moving up a spectrum.

Then you can move to AI or predictive analytics. So I can use machine learning or different models, and there are lots of names underneath the AI bucket, that allow me to do different things with the data.

So I might be able to say, "Based on everybody who renewed last year or bought insurance, based on the price that we charged them, if we were to up and down the price, what would you predict would be the impact on customer retention and sales?"

So now I'm starting to use AI to determine something a lot cleverer, and that allows me to then work out, "Do I need to drop my prices, increase my prices, put more into marketing, put more staff into the business, or not hire as many staff in the first place?"

Barry:  Were the predictions generally quite useful and accurate in your experience?

Kieran:  Very. We all like to think of ourselves as very individualistic, but we all behave to pattern to some degree. It's not saying everyone, but in general, it's true. When I've worked in the past, I could tell with 95% accuracy whether someone would or how someone would react to a price increase or a price decrease.

When you put that into a business, then if I want to earn more money from my customers over their lifetime value, then I can put my prices up for those who are very, as I call it, price insensitive.

So you find this with a lot of insurance companies that maybe put £20 on, £30 on, £40 on. It's almost like, in a horrible phrase, boiling a lobster. You don't notice it, but over time people are earning quite a bit of money out of you.

Insurance, thank goodness, from a legal perspective changed a lot of this, identifying who will or won't take a price increase, and tried to make it a lot fairer.

But you really can see things through data that you might not otherwise do, and that can be highly profitable for businesses. It's why I suggest every year everyone rings around and checks their insurance quote, by the way.

Barry:  Great. Thank you. And back to generative AI?

Kieran:  Yeah, so generative AI is just another form of AI, except there are small differences. So let me talk about broad and narrow AI.

AI tends to be narrow. So a moment ago, I gave you the example that I want to work to build a retention model to determine the price that I'm going to charge a particular customer, to work out whether I can keep them for longer or less. That's a question. You build a team, you get some data, you put a data scientist into it or data analyst, and you can come up with a really good answer. Fantastic.

Now, let's say I want to do something else, which is I want to create a model that tells me whether my staff will stay or leave. In other words, their likelihood over the next 3, 6, 9, 12 months. That's called a churn model in industry terms, and you can do that as well. Again, I build a team, I get a particular piece of data, and off I go. That is narrow because it's a separate use case, and very often it's expensive.

When it comes to generative AI, the first word is important: generation. What generative AI is doing . . . and you will have heard names like ChatGPT or Bard or Bing or OpenAI. All these companies are creating generative AI engines. You give it a particular piece of text, and it's been trained on the web, it's been trained on dictionaries, it's been trained on articles, and you ask it to do something.

So for example, "Can you create me plan for social media, and can you create all of my social media content for the next number of weeks, including articles and posts and everything else?" And it can go and do it.

I could also tell it to create images. So I can go into Bing AI Image Creator and create photorealistic images. I can go in and ask it, "I want you to analyse Tesla. I want you to look at the share price. I want you to look at their last financial report. I want you to tell me whether it's a good investment or bad investment. I want you to tell me how it's doing financially and who its competitors are".

So notice I'm asking this generative AI engine lots of different questions, and I'm asking it to do lots of different things. It can generate all this content and do a whole host of things.

So where one is narrow and expensive, and normally was open to companies with big budgets, ChatGPT is inexpensive. It's free or $20 a month. And there are other platforms that allow you to do this. You don't have to be rich, you can put it on your phone, and you can ask it to do a thousand things. As I said to someone recently, its limitations tend to be your imagination.

But all of a sudden, AI is now available for everyone at a very low price. It's like having an army of interns who can answer clever questions all for next to no money.

Barry:  Kieran, you mentioned ChatGPT a few times there. What is it about ChatGPT that makes it so important, and why is it still almost like the only show in town? It's got such great profile, even so far after it was launched in March of last year.

Kieran:  Yeah. So let me answer the second question first. Normally, whoever moves gets first mover advantage, and then they get the most attention in the world. But this is a truly exciting model.

Generative AI came to the world's attention November 22nd, the year before last, when ChatGPT 3.5 was released. And all of a sudden, people could go in and ask AI questions using natural human language. So in the past, I would have had to write a computer code, and I would have had to use whatever computing language at all to try and get an answer that I wanted.

Now, all of a sudden, I'm able to type in human language words that say, "I want you to go and create this. I want you to answer this question", and it comes up with a really clever, human-like response.

It doesn't mean it's perfect. Later on, we'll talk about hallucinations and some of the problems with the technology. But all of a sudden, in my hand, I have this army of interns who can answer any question under the earth. It can come back with a really great answer, and I can do it typing in a natural way, not in a computer code way.

Now what you're seeing with ChatGPT, and other engines are available, is I can talk to it. That's called multimodal. So I can feed the engine text, I can talk to it, or I can feed it images and get a response.

But as I said, there are other engines out there. From a HR perspective, at your last conference, we looked at a product from DRUID AI. That takes not the world's data, but you can put that on top of your own finance, HR, whatever other data, and then ask it questions. So, "What's the likelihood of my team churning? How many outstanding bills do I have?"

Now you can ask these engines amazing questions to get amazing answers. That saves you a tonne of time.

Now, later on in this talk, and please everyone stay on, I'm going to give you a real-world HR use case example, and I'll play you a video later on of me interacting with that to bring this to life for you. So stay on, and I'll show you that working in the real world.

Barry:  I haven't seen that. Look forward to that. Kieran, just before you start your presentation, can I ask you this? Is it too early to say what sort of levels of efficiency AI can introduce into a typical company?

Kieran:  No, it's not too early, Barry, and this is why people are truly getting excited about this technology. The Boston Consulting Group in the last year or so did an exercise with all of their consultants. Now, to get into the Boston Consulting Group, you need to be pretty special.

They gave everyone ChatGPT access, and what they discovered is that their very, very clever people were able to complete tasks 25% quicker when using ChatGPT compared to those who didn't have the tool. And they also discovered that the work was 40% higher quality, again, compared to those who didn't use the tool.

So by giving people access to this type of technology, you get far quicker completion at far higher quality. And I have to say my using these tools over the last year or two has shown the exact same results. And you see more and more of this data coming out.

So again, from a HR team perspective, they believe, the research analysts, that you can realise 20% to 30% productivity gains now. So you're not waiting. All of this is available now.

That's why this technology is so exciting and getting so much potential. It is literally a wonderful army of interns on your desktop, or in your pocket if you're using your mobile phone, that makes you more productive and your work quality is actually coming out higher as well. Two amazing things.

Barry:  Great. Thank you, Kieran. And before I set you loose for you to start on your presentation, can I just say to everybody at home, if you have any questions as Kieran goes through his presentation this morning, if you'd care to put them into the Q&A box at the side of your screen, then I will try and get to them at the end of the presentation.

But meantime, over to you, Kieran.

Kieran:  Thank you so much. All right, folks, I'll show you that real-world example. When you see it, it blows your mind a little bit.

So Barry's introduced me a little bit earlier on, but look, I've worked in technology and business for lots of years. I love technology, but not for technology's sake. I love what it can actually do. Now I'm working for myself in my own business for the last three years. I'm a lifelong learner and enjoy communicating and explaining all these terms to people so they can get amazing results out of the technology.

If you want to talk to me, get me on LinkedIn. We'll share some of those details later.

But let's dive in. Let's flick to the next slide here and see where we get to today. And as Barry says, ask questions.

I probably don't need to tell you this, but the world is changing. The vocabulary that we're now using is very different than numbers of years ago. And what's driving huge change is technology. It feels like it's the thing more than anything else that's actually moving society and work and everything else on.

So all these terms, some may be very familiar to you. They might not have been three years ago. Driverless vehicles, it seems so natural now, whereas numbers of years ago it wasn't. Intelligent automation, generative AI, hybrid work. Imagine back in the day we didn't work from home. I work from home all the time now, but when I was in business, I did hybrid roles. The oddity is that was always available, the technology was there, but it took a pandemic to get us to embrace it, a burning platform to get change to happen.

But now you're starting to see other new terms starting to appear: digital assistance, conversational AI platforms, ChatGPT, blockchain, IoT. All these things are coming out and they're not new technologies. This is the crazy bit. All of these technologies are here now impacting what we're doing.

Now, how is that relevant to HR and industry? Well, let's flick forward another slide, and then let's start talking about technology and how it's impacting the way we're working.

Today's talk is about AI. Most of that thing that you've seen on the previous screen has AI built into it. And HR is one of those amazing places where AI can be used in lots of different ways. I'm not going to cover all these in detail. The real meat is how do you actually get into using this technology? So let me talk about these briefly.

Again, if anybody has any questions, I write about this extensively. Ask questions at the end or follow up with me on LinkedIn. Of course, I'll answer those for you.

Now, remember, AI identifies patterns. We used some examples earlier on about predicting churn. What it's doing is looking at your past history and determining who, for example, is likely to leave. How do you do that? The odd time it is a case of people start searching on job websites, the performance scores go down, and so on. It can be very hard to see that. And what sometimes happens is that really great employees get missed out in the mix if the companies are too big and we don't notice.

But imagine having AI supporting you. How does this all work? So for example, I look at patterns about behaviours and exits. I feed the AI engine really great data about people's performance and likes and dislikes, their reviews, and everything else, and I can work out who should I be keeping in the business, and importantly, who should I be hiring? In other words, you want to limit the number of mistakes you make when recruiting people.

Well, if I can use AI . . . and there's a great example in psychometrics. If I feed it great information about my existing team, in other words, their likes, their dislikes, the culture of the company, what identifies a really great employee, what identifies someone who maybe isn't particularly great at the job, then what I can do is feed that into my recruitment and hiring process and I can now start to predict a better candidate for my business.

If I actually feed it a little bit more information and say, "Okay, don't just give me a great employee now, but tell me in the future. This is where I want to move in terms of the market", I can feed that data in and then start to use psychometrics and other AI to determine the best candidates.

I can use other AI programmes to on-board them really effectively. Earlier on, I mentioned a particular application, a conversational AI platform and a HR platform at that. It can actually automate a lot of the on-boarding process. So I can send out candidate documents. Candidates usually have lots of answers. They can type into a very sophisticated chatbot, and it will respond.

So all the things that you might be doing as HR professionals, it can provide all this layer of support before the person starts. It can provide this layer of support when they start. So you put the conversational AI platform on top of your HR systems, your finance systems, your recruitment systems.

If I'm an internal HR person, I can use conversational AI to say, "How many days' holidays do I have left?" or ask the company a policy on this or that or the other. What's my performance? What can I do to perform better? Or I can do lots of other stuff around my compensation.

I can actually ask these programmes, "How should I plan the workforce? How many staff do I need in the next little bit of period of time? Who's likely to leave or not leave?"

I can automate engagement surveys going out and the collection of that data coming back in and the scoring of that data, and I can feed that all into my systems to make a very clever HR system indeed.

Any other work I need to do in terms of compliance and risk management and all those forms and documents that go out all of the time, I can automate that.

What does that do? It makes for a cleverer business, but also frees you significant amounts of time to allow you to get on with strategy work, engaging with workers, and everything else that you need to do that's non-routine, for want of a better phrase.

So if you're interested in that, how do you actually put AI into your business? Let me give you the 10 steps on the next slide, and we'll start with a number of these. How do you actually get started with this AI to benefit yourself?

These you can change the order around of, by the way, but they're normally in the right place.

The first thing I do is do what you're doing today, which is spend a little bit of time educating yourself around, "What does AI actually mean? What does it actually do and how can it help my team?" So educate yourself and your rest of your HR colleagues around where I begin, what I want to do, how AI works, where it can make an impact. Is it recruitment? Is it employee engagement? Is it performance management? Where is AI suitable and not suitable? So spend some time doing that, working out what it means and how it can actually help you.

Identify those challenges. Maybe you've got particular parts of your HR function that aren't working as seamless as you want. Maybe churn is actually a particular problem in your business. It may not be. Maybe on-boarding is taking up tonnes of your time. Maybe you want to do something in the future that allows you to be more predictive, more prescriptive, identify better employees.

What I see happening very often is people come up with one use case for AI. And they do the same thing in lots of business areas, by the way. What happens is they then get everybody excited. They have one use case they put in AI, they answer that question, and everybody now wants to do the next thing, but they don't know whatever further opportunities are happening, and programmes tend to run out of steam.

So once you understand what it is you want to do, once you understand how AI works and ChatGPT and all these other tools, you can then apply that to HR and say, "I want to do this, I want to address this challenge, I want to take advantage of this opportunity", and you build out a stream of AI projects. So once you start on one, then you can look at other projects and more projects and more projects to keep the momentum going.

Now, a lot of people go out and get very excited about . . . I call it shiny technology. They see something, they want it, they want to put it into their enterprise. This can be expensive if you go about it in that manner, buying lots of new technology, hiring lots of people.

What you're actually finding today, particularly since the launch of generative AI, is that most technology vendors are actually putting AI into their own systems.

So what I would suggest before you buy any new technology is once you work out what you want to do and how AI can help you, go and look at your existing technology and ask the companies that you're dealing with or ask the IT teams, "Does the current technology actually have AI built in or generative AI functionality built into it, or is it coming soon?"

 You will be amazed how many companies these days are actually putting generative AI into what they're doing. There are very few who are not. So don't spend money yet. Go and talk to your existing provider.

Let's pop on to the next page. So we've got education, we've got the challenges, and now we've got the tooling to match our needs and our wants and our desires. The key one here, as ever, is to be really clear about the business objectives.

Remember, every function, HR, IT, whatever it is, finance, they all serve a bigger purpose. So whatever projects you are putting together from an HR perspective, make sure that aligns with the overall business objectives that are presented to you. If you don't know what those are, go and talk to the CEO, go and talk to whatever department, and say, "What is it that we want to do?"

I suspect and hope most enterprises have HR sitting on the right-hand side of the business, so you're creating the people capability to match the business capability not just now, but for the future.

If you want more staff in your business or more to stay, maybe that's a churn model. If you want to optimise and rationalise your contact centre, that's a different set of AI and business objectives altogether. But again, just anything you do, align it with the business objectives so you're moving in the same direction.

Number five, I say start small and simple, but think big. AI isn't a new technology. It's what I describe as an 85-year-old overnight success. Now, AI is cheap. ChatGPT is free. It's been built into all your applications. This is a technology that works wonderfully well and has done for numbers of years. ChatGPT has just raised its profile again.

But not everybody knows how to deal with AI technology. They maybe don't understand the power of it, the problems of it, the risks, and everything else. So what I always say is to just start small. Start with a proof of concept or a proof of value. Proof of concept means you're trying a technology or a new process. Proof of value means, "Actually, I know the technology works, and therefore we're going to do a small sample".

AI is on the proof of value side. It works. So find a small example where you can introduce this to your business.

Ask a lot of questions. Learn about AI and how it works, where it won't work, the risks, and everything else. Once you've got that in your head, start thinking about everywhere where you can put AI that's appropriate. So start small, think big, and really scale fast, because this technology works.

The right starting point. So you understand how AI works. You're aligned with the business objectives. You're very clear on those. You've got a project to kick off with and you know what you want to do. Now you need to assemble all the various parts.

Don't try and do this alone. You need the right people. Normally, the right people are you and your team who understand HR. So maybe business partners who you're trying to help to use HR to improve their business function or the business overall.

You will definitely need some data and some good quality data as well, and therefore if you can't access it, you might need someone who's in your IT team to help you with the databases.

You will need the tools, and you probably will need some vendor support. You might need some external support as well. So if you or your team haven't got the capabilities and the skills, don't struggle with it. Spend a little bit of money to get an expert consultant to come in to work with you to help you understand all of this technology. And trust me, you'll understand it really, really quickly.

Your IT team will be a wonderful help, and you will need them because it's software technology ultimately and they will provide access to things. But assemble the right team and assemble the right project, and then you're ready to begin on a HR solution.

How do you pick that first solution, though? What I always recommend, go back to number five, is starting small and simple. So this might not be the most impactful project. I tend to do a two-by-two matrix. One is effort and the other is impact. In other words, what effort and cost do I need to put in to get a really great impact out?

So once I've looked at all of the ideas that we could possibly put into place and I've validated those, that they do meet the business objectives, not all of them will be of equal value and not all of them will take the same amount of time.

You might have a really valuable project, but it could take six months to deliver. You might equally have a really valuable project that might take two weeks to deliver. And those do exist, by the way. I would always go after the shorter project to get the win quicker, to build your own knowledge and your own confidence in the business.

But if you can work out all of the HR case studies, all of the example places where this could work, and put them all into this table having looked at, "What's it going to cost and what's the benefit I'm going to get? How long is it going to take?" then that's a really nice and orchestrated and structured way to determine Project 1, Project 2, Project 3.

You'll want to build a business case for this, because it will result in some expense, but also the benefits will be there as well. Ideally, do all the projects that get you more benefit than cost you more money, and you'll end up in a really great place.

Now you've worked out what you want to do, now you've got the team assembled, now you've got the technology and tooling ready to kick in that first prioritised use case, and you start to implement the AI. And that could be anything: a new better recruitment, a new on-boarding process, a conversational AI tool that allows you to ask HR lots of questions. Train your HR team how to use the new tooling.

All the time you're getting this done, don't have it done to you. In other words, hand it to your IT team and they come back in a couple of months' time and say, "There you go". That never works. You want to involve yourself in the project, you want to ask lots of questions, and you want to make sure that everyone in the HR team is trained to use the tooling and technology.

Too many times I've seen brilliant technology but no training, and of course the technology rollout and the implementation and the use fail. People spend a lot of money, but don't end up with the benefit that they've got. So do train your team and yourself how to use these tools.

Once you've implemented the tools and you've started in this pilot or proof of value, keep monitoring it. Don't do it and walk away, because what happens is life changes. The results coming out of the model might not be perfect day one, but day two and day three and day four they'll get better. If you constantly look at it, constantly refine it, constantly fine-tune the answers, you can have a really amazing system.

Like all of us, we all need time and attention a little bit of training. AI and AI programmes are exactly the same. So once you know how to use it, you're now monitoring and evaluating the tools or the performance of your model, and you're now refining it constantly to get better and better and better answers, just like you would train someone. Start to now expand the platform and the use of the tooling.

Remember, start small, learn, move in simple steps, do everything you can to train yourself to do it, and now expand AI across the enterprise.

There are lots and lots of examples of where AI can be used. I'll put up some information around 10 or 15 different use cases. But trust me, AI is only limited by our imagination. Where you have something that involves a judgement and is reasonably routine, you can put AI in and get tremendous benefit out of it.

Those are my 10 steps. Let me flick on one here as well. I promised you that I would give you an example of all of this work, and I'm going to use the AI model that is ChatGPT. Now, there are other models. There are other tools that are available.

I was approached a little while ago by a company in Northern Ireland, or a group in Northern Ireland, and they have an HR function. And what they wanted was, "We're really struggling when it comes to our applicants and our interviews". And what I did was I said, "Okay, give me one of your job descriptions. Tell me how you go about actually constructing an interview. What are your values of your business? What is your interview methodology that you want to use to find the best candidate? And what is the scoring mechanism that you want to put in place to get all of this seamless and perfect, and how long does it take?" They were saying, "Well, this usually takes about three or four days per candidate per role".

Let me show you what I did in five minutes or less using ChatGPT. So if you wouldn't mind, just hit press play.

Video: Okay, folks, let's introduce you to the power of ChatGPT. Let me move myself out of the way. Let me show you ChatGPT. But do remember, this is only one particular tool. There are lots of them out there.

Before I begin, let me do one thing. Let me go down to Settings and Data, let me go down to Data Control, and let me turn off Chat History. What this does is secures what you're trying to do. In other words, if you did not have this box checked, anything that you drag into this window suddenly becomes part of the corpus that is ChatGPT, and potentially other people can access it.

So I've turned that off. I'm not in a private session. It won't remember what I'm doing.

What am I going to do? Well, what I'm going to do is feed in the senior HR adviser role by just dragging that file across. This is your PDF, and now I'm going to tell ChatGPT what it is.

Barry:  Can I just say, Maria, I don't think we can hear or see this video. Maria, is it possible just to try a few more different buttons? Because I'm getting quite a bit of Q&A in saying that they can't actually hear the video.

Maybe we can come back to this, Kieran, just whilst Maria attends to this. Sorry about the technical issue there, but that's what I'm getting. So apologies there and apologies to everybody, but let's just continue for a moment and see if we can get back to that.

Apologies, Kieran, but if you could carry on with your presentation, we'll see if we can get back to that.

Kieran:  Yeah, someone said, "Never work with children, animals, robots, or video technology". The worst case, if we can't get people to look at this, Barry, what we'll do is send it out to them.

But what you were going to see there, and we will come back round to it, is I took the job description . . . And again, notice I've been very cautious at the top end turning that box off that says, "Don't make my data part of ChatGPT". A number of companies actually fed their data into that system, didn't click that box, so all of a sudden now private and confidential information becomes part of ChatGPT's answers or training set for other companies. Please don't do that.

But if we can get back to that in two seconds, I'll explain what I was doing, and we'll send out the video afterwards.

Look, once you've worked out what it is you want to do and you've walked through those steps, and you're able to support a really great AI system, you'll still have on-going questions to answer overall. These are the summary questions.

Get your IT team's support. They control, to a certain degree, the keys of the kingdom, your access, your applications, the databases, access to the data itself that you'll need for all of your systems.

They're also really supportive when you involve IT teams up from the front. What I've seen too many times is people not doing that. Involve your IT team, get their support, and they will work with you to allow you to get all accesses and things that you need, and the backups done, and the security taken care of, and the HR and the AI policies adjusted to allow you to implement this technology.

This type of ambition, you can have a bright and shiny picture, but you need to be able to paint this vision to other people. So remember I mentioned earlier on about actually understanding the technology? You're going to get asked, "Well, why should we do it? Why should I engage with them? Is this actually working? Is it giving me the right answers?' The more you can have a bright and shiny vision, fantastic. You should, and AI and generative AI should be part of that. But understand the technology to be able to answer all the questions that people have.

There are too many scare stories out there. "AI is going to replace jobs. Generative AI is going to decimate things". Obama in his exit speech said AI is going to kill lots of American jobs. J.F. Kennedy did the same thing, by the way, 60 years ago. That's not what happens. Put this technology in to augment really great people.

There are routine things that we do when you break your job function down. I gave the example of answering lots of what I describe as very basic questions. What is my holiday entitlement? How much money do I get for this? What is the policy for that? You can answer those all day long, but you can actually get conversational agents to do it. And that's a very basic example that's working.

But all this takes time, talent, and funding. So you're not going to do this for free, but it's reasonably inexpensive as well. Again, get the talent and your knowledge up. Bring in the support that you need. Recognise it's going to take a little bit of time, so put some programme management around this, put some proper funding around it as well, and you can come out with amazing things once you've learned what to do.

The other bit here is a lot of companies at the moment are going, "Will I wait? I don't think this is going to impact me". I call that the will and the skill element. AI is everywhere. I've been in technology for 30 years, and I have seen not just the excitement . . .

Remember, all technology really rises up in what I call the hype curve. So you remember a couple of years ago, metaverse was going to be huge. Facebook changed its name. It may still be huge. I can't really see that. Blockchain, Bitcoin, all of these things were absolutely massive. They go through hype cycles. Again, I couldn't get that one.

I've worked in data science for 13 years, leading data science teams. I watch technology trends. This is one that isn't going to go away. Watch what Microsoft are doing building in Copilot, their generative AI engine, into everything they're doing. Every one of the big manufacturers, Canva, pick your partner, pick your technology, are all putting AI into everything. Why? Because it works, and because it's affordable.

This isn't going away. If you're not going to do it, your competitors are already doing this. So please, please, please, have the will and have the skill. You need to use AI. It is going to impact how we operate. It is going to impact jobs. Get ready and get prepared. No, I don't mean that as a frightening statement. I mean that as a call to action, and everything will be wonderfully good in the world.

The other bit here, and this is more a statement, is "Are you in a constant state of reinvention?" Let me explain that. Technology is impacting what we're doing. Things have never been faster, and things will never be slower because technology has that type of driving, driving, driving impact.

As Barry mentioned at the top end, 12 months ago, we weren't even talking about ChatGPT. Now every dollar that I can see from venture capitalist companies, from big investment firms, is going into AI because this technology works. That's going to accelerate how much technology is actually impacting our roles and our lives and our businesses and our markets around us.

So what we're going to have to get used to is constantly being in a state of change, constantly watching what's happening and how technology is driving the market, and constantly amending how we go about what we deliver, how we use AI to deliver that, and how we get folks inside of our business, including ourselves in HR, the skills that they need to be able to have a wonderful career with us for as long as possible by constantly keeping an eye on the market and training and retraining and investing in new ways of working, in new technologies, particularly AI, conversational AI, ChatGPT, and others.

Barry, I'll come back to you now to see if we can go to the video. And if we can't, I'll explain what it is, and then we can pass that video out to other audience members as needed.

Barry:  I think, Kieran, what we'll probably do is send the video out to everybody after this, simply because we've still got quite a bit to get through. We are down for just 45 minutes, and there's quite a lot of Q&A in. So if that's okay, that's the way we can go.

Kieran:   Absolutely. The video is self-explanatory, but if anyone struggles, they can come back to us.

Barry:  Okay. Do you want to finish off, or shall . . .

Kieran:  All done now. We can go to Q&A.

Barry:  Lovely. Just before we do that, can I ask Maria just to advance the slide to the slide about 6 March? I just wanted to mention to everybody that we have a full-day event coming up online on 6 March, and it's entitled "The Fundamentals of AI for HR".

Delighted to say we've got Kieran, who is our keynote speaker for that event. We have loads of organisations already booked for it. We've got some really great speakers tackling some very interesting topics, which I know will be immediately useful to those in HR.

So if you haven't booked on that yet, please do so because we think that's going to be a great event, and it's going to be really useful to everyone who attends that.

Kieran, just to go to some questions here. We've had some really great questions in, so thank you to everybody who has sent things in. Question here from Matthew Hamilton, who said, "Kieran, how do you overcome the voices of resistance who always call out data privacy and data security concerns/risks around the use of AI tools? ChatGPT is a prime example when it was launched. So many organisations have approached the adoption of AI tools with almost a blanket ban". Thank you, Matthew. Great question.

Kieran:  That's a brilliant question. I love it. I sometimes think there are two mind-sets, Barry. There's pain and there's gain. The pain mind-set says, "Oh my goodness, this isn't going to work. This is problematic. There are privacy concerns. There are responsibility concerns. We're all in trouble".

There are risks in everything in life. There really is. There's risk stepping out the front door. Remember, earlier on I said to learn more about AI, learn more about the tools, learn more about being able to respond to that type of question so that you project confidence with people and you're able to explain it. That helps you overcome resistance.

Some people never change. That's unfortunate. And therefore, you're just going to have to push forward. AI is going to literally take over everything, so you really need to do that.

Everything comes at a risk, so let's work through some of the risks. One of them is data privacy. ChatGPT, by the way, by default, if you copy something into it, you're giving ChatGPT permission to use that information as part of its training set to allow it to produce better answers for everyone else. By simply clicking that button that I showed you earlier on, all of a sudden, you remove that risk.

What they are doing, OpenAI, as the manufacturer of ChatGPT, is now starting to introduce a business version of this where your IT team has far greater control over access and data permissions and everything else.

If you don't use ChatGPT and you use other tools . . . Microsoft Azure have a generative engine. It comes as a really secure engine out of the box. Your applications that have generative AI built into them, that doesn't allow data out of your network.

So again, there are some risks with this. They aren't really new, to be absolutely transparent, and they can all be overcome with a mixture of controls and permissions with some usual policies.

And the key one for me is to always train your teams to what the risks are and what to look out for. It doesn't matter what that is, be it generative AI or something else. Once they know all those things, they can learn to avoid them or learn to cope with them.

Barry:  Okay, Kieran. Thank you. A question that was in earlier, and I think maybe you just covered it in that answer there, is, "Are the data protection considerations concerning the data that can be fed to ChatGPT or other generative AI servers?"

And also, Carmel Murphy has just emailed in and said, "Sorry, I missed the part where you showed the button to remove chat history and disable the training function. Could you remind us how to do that?"

Maybe we could go to that right at the very end as just a sort of a repeat bit. But did you want to talk about the data protection considerations and then another question?

Kieran:  So let me answer this in two parts. There are a lot of conversations at the moment around, "How are these engines trained?" And some of the vendors, like OpenAI, are now being sued by a variety of authors and by a variety of publishers, big press agencies, because they're saying, "Look, you took our IP, you fed it into your systems, and you're now using that to train your system to come up with better answers".

When you dig really behind it, what these engines have done is taken 700 billion documents and articles and data points, and now they're just using prediction to say, "I've seen this word following this word a thousand times. I've got all this data. I can now create something new".

So they're suing because possibly, probably, they've used data that belongs to someone else. It'll be interesting to watch the outcome of that court cases. So let's see. That's slightly separate.

When it comes to your IP, which is a lot more important, you do have to be cautious. So for example, if you're writing letters to MPs, if you've got someone's performance evaluation, if someone has sent you in a CV, what you could do is drag it into the ChatGPT window or other tools and you could say, "Give me a summary of this application. Tell me an email reply that I should send back to this question", and so on.

But what you will have done if you haven't ticked that box, and it's at the very start of that video, is you have said, "You can take my data and you can use it for whatever purpose you want". Please tick that box to stop it being part of OpenAI's system, if you use that one.

Engage your IT team. It's probably them who are introducing the policies in the first place. They're slightly cautious and nervous about this, but work through these things, because you've had to work through them before.

If you have an HR application sitting in the cloud, that's been locked down, secured, everybody has gone through the protocols, and people have been trained what to do and not what to do. This is just another technology and you just need to go through the same iterative process. But please do it quickly because this technology is coming like a hurtling train.

Barry:  Okay. Thank you, Kieran. Probably the last question. If you're working with your IT team to introduce AI in your organisation, what key measures would you include so you don't end up with something that isn't fit for purpose?

Kieran:  Let me use an HR example and come up with something that I hope aligns with your business objective. Every year, you probably will be given KPIs. You'll be given goals and targets to the HR team. Let's imagine you run a contact centre and your contact centre suffers from a high attrition, high turnover of staff. So what can you do? Let's work out what that's costing you for years.

In a company I worked in the past, I went to the accounts team and said, "Every so many months, we're losing 20% of our staff. That's costing us money". What did I mean? Well, we needed managers off the shop floor to train these people. We needed a lot of HR people and a lot of job adverts that cost a lot of money to recruit in the first place.

Then we had people outside of the business who were interviewing these people. Then we had more risk and compliance people because we needed to check their work more often. Then we had rework because, in fairness to them, they weren't trained, they were doing the work, they made mistakes, and we had to correct that. Then we had customer attrition because this was a volume business where we were selling product and so on and so on. All those things cost.

What we did is we did an AI project, we looked at the reasons why people were going, we looked at the type of character an individual that we wanted in the business, we fed data into the model to work out the best characteristics, we put that into a psychometric profiling tool at the beginning, and we told our interviewers what to look out for.

We adjusted our interview questions to make sure we were asking questions that got us a better candidate who was more resilient, who was more in tune with the values of the company.

And this is back in the day, by the way. The six-month cost of every time we lost someone was £29,000. We moved attrition from just under 30%, and if I remember right it was about 27% to 29%, to 5%.

So if you can work with your financial department to work out, "If I drop attrition from whatever it is per cent down to this per cent, how much money will be saved per year?" Get that number, and then work out what it's going to cost to put an AI project in place that allows you to do exactly what I've described. That's a really good way of having solid hard numbers around every project that you do.

Barry:  Thank you for that, Kieran. And just before we wrap up, I just want to say this to everybody watching or listening here. Kieran is a very busy man. Each time I reach out to him, he's usually flying somewhere to do a keynote on AI or to chair an AI conference, be it in Amsterdam, London, or somewhere as fancy as . . . Was it Mauritius you were in recently?

Kieran:   A 40-hour flight, 30 hours on the island.

Barry:  Not so nice, is it? But I bothered Kieran into agreeing to give us a bit of his time free of charge. What he has kindly agreed to do is to offer his calendar out to the first five people who want to book him for a complimentary 30-minute one-to-one chat about their particular AI needs in the workplace.

If you want to avail of this opportunity, then I would advise you to click on the link that you see in front of you as soon as possible, possibly even right now, because it will be limited to just five people.

That is a very generous donation of your time there, Kieran, so thank you for that.

We will be following up with an email to everybody, which will include a link to the video and also to your diary as well.

But can I just thank you there, Kieran, for a fascinating talk about the application of AI to HR. I agree with you. I think it's a fascinating area. There's loads of opportunity for organisations, for HR, for HR people to really embrace AI and to run with it. I think it's absolutely fascinating indeed.

So once again, Kieran, thank you. Really appreciate your time and your presentation this morning. Thank you to everybody attending, and I look forward to seeing you again very soon. Thank you.

Sponsored by: