Will AI finally understand your feelings?

Could the future of customer service and workplace harmony lie in the realm of Emotion AI? Join us for an enlightening conversation with Marie Toft, CEO of Emotionise AI, as she unveils the transformative power of weaving emotional intelligence into artificial intelligence.

Marie's unique journey from psychotherapy and broadcasting to spearheading a pioneering AI company sets the stage for a discussion on how Emotion AI is bridging the emotional chasm left by traditional AI systems.

This episode is packed with insights into how Emotionise AI pre-emptively tackles communication hiccups, enriching both customer and employee interactions by fostering connection and understanding.

We explore how emotionally aware AI can enhance customer service outcomes, the challenges of integrating emotional intelligence into AI technology, and the future opportunities that agentic AI presents while highlighting the critical aspects of trust and ethical considerations.

• Definition and significance of emotional AI
• Importance of emotional intelligence in customer interactions
• Challenges faced when building emotionally intelligent AI
• The role of data quality in developing effective AI models
• Benefits of agentic AI for personalized customer support
• Importance of establishing trust in AI technologies
• Future implications for emotional AI and business operations

We also venture further into the implications of Emotion AI for business success, highlighting tangible outcomes like improved customer satisfaction and a potential 25% profit boost. Discover how creating emotionally intelligent AI models aligned with company values can significantly impact a brand's bottom line.

We also explore the role of senior leadership and the necessity of clear metrics such as CSAT and NPS for evaluating AI initiatives.

Marie Toft shares her insights on the broader importance of emotional intelligence in both personal and professional spheres, offering listeners a chance to connect with her for continued learning.

Tune in to unlock the potential of emotional intelligence to revolutionize your business landscape.

See: Emotionise AI for more details.

Chapters

• 0:00 Emotion AI in Technology Industry
• 11:20 Implementing Emotional AI for Business Success
• 23:19 Connecting Emotionize With Emotional Intelligence

Transcript

Speaker 1: 0:00

Have you ever used an AI chatbot or visual AI and felt that something was missing or not quite right? Did you get the feeling that AI lacked emotion? Well, now we have an answer to these problems. Emotion AI, also known as effective AI, is a subset of artificial intelligence that measures, understands, stimulates and reacts to human emotions. Emotional AI helps machines gain an understanding of human emotions and can be used to develop real empathy. If there is such a thing. Welcome to AI Unscripted, where I talk to the creme de la creme of the world's best technology companies. And today I have Marie, CEO of Emotionize AI, an Irish-based global provider of agentic AI solutions that has successfully harnessed the power of emotional intelligence with trained AI. Welcome, Marie, Marie. For those who don't know you, would you mind giving them a brief introduction to you and your company?

Speaker 2: 0:55

Thanks, ciarán, and thank you so much for the invitation to be on, much appreciated. So Marie Toft is my name. I am the CEO and co-founder of Emotionize AI. So to emotionize something is to make it more emotional. So we are a generative AI company. What we do, as you rightly point out, is we have harnessed the power skill of emotional intelligence to our human trained AI. And where are we basically finding traction? What we do is we help improve the outcomes of customer support, particularly, and employee engagement through written communication. So, basically, if you're writing an email or at town hall, or you're responding to a customer who's having difficulties, our AI is an ambient, intuitive AI that comes in and gives you feedback on your communication, saying look, are you getting this? This customer seems to be pretty upset. Perhaps you know this is the way you might respond instead, and then it will offer you a more emotionally intelligent replacement sentence.

Speaker 1: 1:57

Wow, wow. So, marie, what's driving this trend to adopt emotional AI or AI agents in the industry?

Speaker 2: 2:06

Emotional intelligence is fascinating. So, just to give you a little bit of background, I'm a trained psychotherapist. I'm also I had a 20-year career in broadcasting as well, before I became found myself in the unusual situation of being CEO of a Gen AI company, and so what we've done with our model is we've trained it on both of those skills very good communication skills armed with emotional intelligence. And so, when it comes to emotional intelligence, when I went back to study psychotherapy, I discovered emotions are vital. They are the barometer in which we navigate our world and we ignore them at our peril and we make decisions based on our feelings, and we need to be alert to that. So, basically, emotional intelligence also is in quite short supply, and people are always shocked by the statistic that I'm about to tell you, ciarán, but only 36% of us are emotionally intelligent, and people go really, and then everybody looks around the room and thinks am I one of the 36%?

Speaker 1: 3:06

Or they think you two aren't.

Speaker 2: 3:07

but I am Absolutely because actually that's a really good point. We all believe we're more emotionally intelligent than we really are, even the 36% of us. What do I mean by emotional intelligence? We work with our model. We have three factors of emotional intelligence.

Speaker 2: 3:22

To be emotional intelligent, first of all, you need to be self-aware. Unfortunately, that's in quite short supply. To also be emotionally intelligent, you need to have the ability to regulate your emotions. So if you're under pressure, if you're under stress, if someone's coming at you, you don't respond in an angry, harassed, difficult way. You can stand back, take a breath, take a pause and respond in a more diplomatic way Again, not a skill, unfortunately, that is in, you know, large supply. And finally, emotional intelligence is vital. It helps us to navigate relationships and it certainly helps to improve relationships.

Speaker 2: 3:57

So, if you think about it, let's talk about customer service, particularly at the moment, good customer service is completely reliant on good recruitment and good training, and you know there's a lot of investment going into that and it will certainly help. But we're all only human and we all have bad days and unfortunately we a lot of large customer support agencies are only getting the problem after it's happened. It's retrospective. Someone's having a bad day, they might have had a fight with their teenager that morning, and you know what the customer gets it, and then you're moving in to solve a problem that's already happened, Whereas with our model we're there, we're ambient, we're intuitive, we're moving in, we're live, we're dynamic, before the problem even begins To say to, let's say, the customer support agent look, this customer is really not feeling particularly great about it.

Speaker 1: 4:49

We suggest maybe you tone down the language here or you know, respond in a more emotionally intelligent way and assuming they can identify that afterwards. So what is an emotional AI agent and how difficult is it to build emotion into AI?

Speaker 2: 5:01

Well, there's two things happening there. Let's talk, first of all, about how difficult it is to build emotion into AI. So, in 2020, I decided to train AI to be emotionally intelligent and everybody thought I was mental. And I saw GPT-3 in 2020, ciarán, and, if you well know, gpt-3 was the forerunner of ChatGPT, which was then launched, you know, well, over, you know, nearly two years later, over two years later. So I saw it then and I thought right, I'll be able to build this now. And so how did we do it? We so cause. We were one of the first Irish startups to see generative AI, to see where large language models were going, and we were one of the first Irish startups to actually begin working with open AI. We started working with OpenAI in September of 21, well over a year before ChatGPT was released, and what I discovered and this is still true of all large language models, ciarán how to build emotion into AI is data.

Speaker 2: 5:56

Everything is data. When it comes to large language models, these models are only as good as the data on which they are trained, and so, over the last few years, we have really fine-tuned and finessed how we create that data, and the key to creating that data are people. So we hire writers and we test them on their communication skills, because when you're communicating through written text, emotional intelligence is all about the language that you choose. If you choose the right language, you will connect emotionally with your audience. So they need good language skills, good vocabulary skills, but they also need it. Certainly they need to be in that 36 percent of us who are emotionally intelligent. So we writers come to write for us and they help us label the data and we actually this is not scientific, but only 30 percent of the people who apply to write for us pass those communication and emotional intelligence tests that we give them, which is actually eerily similar to the 36% of us that are emotionally intelligent. So they label the data and it's through the creation of that data, that high quality data, that we have trained AI to be emotionally intelligent.

Speaker 2: 7:04

So the second part of your question are AI agents, and I know you've spoken a lot about this and everybody's very excited about AI agents, and this is the year that agentic AI really takes off. So agentic AI is genuinely fascinating. We have been using agentic AI since January of 24. So we were again a real pioneer in that. There's two things that's really helped us with. The first thing is what agentic AI has done for us particularly is it's allowed, it's really improved our accuracy and our quality. So any of your audience who are using large language models, there's still a bit to go on accuracy and quality and in many ways, as you well know, it's sometimes it puts people off using it because it doesn't always get things right. Now that accuracy has improved phenomenally over the last year, particularly. But agentic AI.

Speaker 2: 7:56

With agentic AI, what you do is you train the agents to be very, in a very, very specific task. So, for example, we've now trained an AI agent to read an email thread and understand the context and the emotional temperature of that thread. We've trained another AI agent to populate, to automatically create an emotionally intelligent response. So one AI agent is doing this, one AI agent is doing this, and then they're talking to each other. And that's a really important part of agentic AI, because when you have one, before this it was a bit of a blob of AI and it got a lot of it right, but it got some of it wrong, whereas now you're tailoring every agent to do very specific tasks, they're talking to each other and what we've seen is a really big increase for our model. By the way, we call our model EMER, the, the Irish girl's name, because the first three letters stand for Emotionally Intelligent Model. So we have seen a really, really big improvement for us, certainly in both quality and accuracy, using agentic AI.

Speaker 1: 8:55

Oh, wow, let me pick some of that apart there, because if I'm one of those talented writers and you've just told me I'm not emotionally intelligent, I could have a depth, it could have a devastating impact. But as people, we like to think that we're not logical to a degree and so prescriptive and linear and something else. But you know, we we kind of are, and you can build formulas into us and build formulas into people which we never like to hear about. And that piece there where we're saying you know, I suppose, marie, we're very forgiving of people who make mistakes, who maybe aren't emotionally intelligent or make the blob a little bit squidgy at times, but we're very unforgiving of AI. So what are some of the pitfalls that organizations feel they're encountering when they're attempting to integrate emotional AI solutions into their business workflows? And are they being a little bit too judgmental of AI where they're not as judgmental, critical or willing to give up on their human workforce?

Speaker 2: 10:04

Really good point. I think the whole accuracy and trust part is huge, ciarán, actually, and I think that's really, really important. What I've discovered as we roll Eimear out in workplaces is they're I find generative AI particularly really interesting. It's a complete paradigm shift because, for the first time, humans and AI are evolving together. We said we call ourselves a cognitive collaborator. We're there as a kind of a safety net you can take on board, and we're also human in the loop. So we give you reasons also, which is really important. This is even from a learning and development point of view. We give you reasons why this communication may not be appropriate for the person. You know, for your target audience. It's up to the person whether they want to take on board that feedback or not. You know, sometimes they know that maybe this they need to be a little bit more more firm, let's say, for example, with this customer.

Speaker 2: 11:02

By the way, the leaps in ai we're already seeing is ai is eventually going to know the whole history of that customer, where you were with them, and the ai agent will be doing that, and so what we're discovering, though, is I find the whole element because one of the things I also do as well, as you know, run emotionize ai is. I'm also and this is based on my own experience of rolling ai out to you know, to organizations and workforces there's three things going on there's a huge education gap, there's a huge communications gap, and what that means is there's a huge trust gap, and so I read a really interesting report this week about you know where we're going with ai in 2025 and it's it appears limitless, doesn't it this this year? You know where we're going with AI in 2025 and it's it appears limitless, doesn't it this this year? You know the first couple of weeks of 25 alone has been, you know, got Sam Altman saying AGI is much, you know, closer than we think and it's just moving so quickly.

Speaker 2: 11:56

But the limit to AI is actually trust. We don't get trust right. We're leaving swathes, millions of people behind us in where we're going, and you and I are very plugged into ai. It's our living, it's what we deal with every single day. But I don't know about you when I go out to organizations, there's people there who have never even used chat, gpt, and what I find when you roll these models and ai out to organizations, if you don't get the education and communications and trust right and, as you very rightly say, it's that trust element. Actually, when it comes to is it getting it right? Is it not getting it right? So it's beholden on startups like me to constantly improve quality and accuracy there's nothing that matters more but also to educate people about this paradigm shift.

Speaker 1: 12:44

Yeah, that. No, I agree wholeheartedly. I come into that. I think I see a mixture of, you know, fear. Number one oh my god, it's ai. It's going to ruin the world and we've got agi and, my goodness, this is. You know, humans are lost, we're tragically a waste of time and the machines will take over to fear. I don't know what it is and therefore I'm going to bury my head in the sand, but you really have a case of what I describe as digital darwinism.

Speaker 1: 13:08

You know, if you don't actually invest in this, you really really are in trouble, and there's too much money going into it for this not to be something and you and I've been in technology and business for 20, 30 years we we know the trends that won't work and we know the things that will and this is a real will work.

Speaker 1: 13:22

But if I'm a company, so Marie, just to move us back a little bit. If I'm a company, it's going to take money, it's going to take time, it's going to take investment, education and tech, and you know all the things that make a successful business strategy. So what are some of the components you know for businesses to implement a successful emotional AI strategy, to get, dare I say, a return on investment using AI and agents and everything else in between?

Speaker 2: 13:51

Well, let me tell you, I suppose, about our approach, particularly because that'll kind of give you an idea when it comes. First of all, you need to understand why emotional intelligence matters. Some people are on that train, some people aren't. I mean, harvard has shown that if you connect emotionally with your customer, your profits can increase by up to 25%. And, what's really interesting, we spoke to a very large company last year and they had very interesting experience of this. So they're a technology company and when customers come to them, they have two things going on They've got their tech problem and they've got their personal problem. And their research has shown that sometimes the customer service agents would solve the tech problems very quickly, but not the personal problems, and their NPS and CSAT scores were low. But sometimes they couldn't solve the technical problem straight away, it was a bit more sticky. But they would solve the personal problem. They make sure the person felt valued and heard, and their nps and c-sat scores were high.

Speaker 2: 14:45

So emotional intelligence matters and it moves the. It is not woo woo, it is not, you know, woolly, it is actually a power skill and what we've done is we've democratized it. So when it comes to implementing how this, this technology. Again, like I said, kieran, it's back to data. What we also do, as well as emotional intelligence, is we can train our models. We create bespoke models in your brand, in your values, in your tone of voice. And this is huge for companies because at the moment, they're relying on style guides and training guides and as human beings, we can't hold all of that information. So the way you communicate if you were working, let's say, for Virgin Atlantic, would be very different to the way you communicate if you were working, let's say, for Virgin Atlantic, would be very different to the way you communicate if you're if you're working for Barclays Bank. And we can get that tone and style. How do we do it? Small amounts of data. We work with companies. We only need small amounts of their data, and then our team and our expertise labels that data and we create a bespoke model.

Speaker 2: 15:40

But what I'm finding that's not a difficult. It is a difficult thing to do. It's taken us four years to perfect it, but in terms of time scale and so on, we can do that within about six weeks. It's not massive because that's our expertise, that's what we do, that we, that's what we spent years perfecting, and we know how to create emotionally intelligent AI.

Speaker 2: 16:00

Here, what I find really interesting is the implementation of that model. That's where the rubber hits the road and what I'm finding again back to what I was saying earlier is the buy-in of senior leadership, the buy-in of everything to come from the top down the education, the communications we have, companies create success metrics, kpis, is it CSAT, is it NPS? And all of that piece of work needs to be done and really it's the start of their AI journey and it's that experimentation, learning, realizing the AI is just going to get better and better the more you use it. So that's the journey when, if you want to embrace emotional AI, that's there the stages of the journey you need to go on. Emotional AI that's there, at the stages of the journey, you need to go on.

Speaker 1: 16:43

Yeah, and I suppose that it's that you know. As you say, I'm kind of excited by this In that when I've ran and built contact centres and operation centres In the past, it was that hiring and recruiting of people that you know For want of a better phrase Weeding out over a period of time the better, the worse and whatever else, and then the lack of consistency and quality and the retraining and the fixing and everything else. So augmenting people, you know, with great, emotionally intelligent answers that give a great technical answer, give a great emotional answer, and having that consistency time and time again that reinforces the brand and encourages customers to come back and back and back, is really exciting, you know, and, as you say, it's the old phrase, almost trite now, but so so beautifully put together. You know, ai is as bad as it's ever going to be and it's pretty bloody amazing at the moment. That's the extraordinary bit. So what do you see as the future of emotional AI, marie? What do you see as the future of agentic AI?

Speaker 2: 17:39

agentic AI and emotions is going to and we are already doing this, kieran. What agentic AI is allowing us to do is actually create mass personalization at scale. So this is where agentic AI is really, really powerful. So, if you are an 80-year-old woman whose broadband has issues and you're in Roscommon, and you're in Roscommon, the language that a customer support agent will use to help you is different to if you are a 15-year-old skateboarder whose wheel has come off and you're trying to get in touch with the company to say what's going on, and so the language that you use, the tone that you use, would be different. And again, as humans, it's hard to keep you know, to switch like that in our heads, and even you know. So agentic AI is moving in. It's saying you're now talking to Joan, you know, maybe this is the language you should use, or this is the third time she's contacted you, or whatever it is. So, agentic AI, what you're going to see is mass personalization, and then the emotional part. That's fascinating. That's only going to get bigger, and sometimes I find, when I speak about this, ciarán, people get very nervous about the idea that AI has, you know, the ability to replicate emotional intelligence, and we need to be careful of the language that we use. Ai is not emotionally intelligent. It is trained to find ways of you know. That training process is what we've perfected.

Speaker 2: 19:08

But I always find the whole AI journey, as I've been witnessing it over the last few years, we didn't think it would, it could replicate creativity and we didn't think it could replicate empathy and emotional intelligence, and what's fascinating is they've been kind of growth areas. Actually, this is the year where emotional intelligence really comes to the fore in AI and we're delighted to say we're ahead of that curve, but this is going to get much bigger. Ciarán, in terms of honestly, as a psychotherapist, as a psychotherapist, even 10 years ago, they discovered US veterans returning from the wars in the Gulf found it easier to speak to robots when it came to PTSD than people because they felt less judged, and I found that really, really interesting. And we're going to find there's going to be emotional AI there to help you and that's going to be an absolute game changer and we're going to have to really think about what that means.

Speaker 2: 20:05

The big thing and I know you talk a lot about this is regulation, data, data security, data privacy, because we're now revealing parts of ourselves that are really what it means to be human and what it means to navigate this life, and I think that is actually where we're headed. But we have some really important ethical and regulatory decisions and, you know, things that we have to think about as we go down. And, as you're very well aware, the technology is completely outstripping the legislation, but this is quite serious stuff and we do need to really take a pause and go well, where are we heading with all of this now?

Speaker 1: 20:44

Not to a degree. I absolutely fundamentally agree with absolutely everything you said there, and not for folks who are listening in, should that put you off? This is just natural and normal when it comes to all of the technology. And it's interesting, marie, but you mentioned a moment ago that you know AI isn't emotionally intelligent. Then, dare I say it, let me say it, not you, nor half the people that I meet on a regular basis, and therefore, using something like AI to assist and improve people's emotional intelligence and company's emotional intelligence and empathy and everything else, feels like the right way ahead to do.

Speaker 1: 21:17

And I always find it fascinating, because how do you get an amazing agent? Well, hopefully, you train them in the first, or you find them and then you train them, and then we, as managers or as leaders, are human, prompt engineers. So the fact that you're training an AI to do the same thing or perform in the same way, but consistently and persistently, and the benefit of a 24-7 is absolutely, you know, a real tick in the box. It's something that companies should be doing, should be investing in, to provide that much better experience, and I have seen that time and time again the better experience that a customer has, the more they're likely to persist and stay with your brand, the more they're likely to spend more money with you and the more likely they are to recommend.

Speaker 1: 21:59

And now the fact that customers globally, with the point and click of a button, can move between you and someone else, then you need all of the technology at your fingertips to ensure that you're the one they're not moving away from and that you're available for them 24-7 to be that empathetic organization that they wish they always had. Marie, for those who don't know where to find, emotionalize AI. How do they find out a little bit more about you and about the company to be able to take advantage of what it is you have to offer?

Speaker 2: 22:37

of course, kieran. And, by the way, we'd be delighted to give um. We always give live demos, so if anyone is interested in in contacting me, we'd be delighted to show you how how emir works. So emotionize ai wwwemotionizeai is our website and we have a lot of information, particularly on the science, and you know how this matters and um without you know, obviously revealing our IP and our secret sauce. You know why human trained data, and I couldn't agree with you more.

Speaker 2: 23:04

What we've effectively done is we've democratized the power, skill of emotional intelligence. We've made it available to a much wider audience and I'm really passionate about it because we need to make AI human more rather than human less if we're going to embrace it. Human more rather than human less if we're going to embrace it. And linkedin actually, if anyone wants to connect with me on linkedin, marie toft, um, you know, ceo and co-founder of emotionize. Again, like I said, I, as well as the, the, the startup um, I do some keynote speaking on all of this as well and I'd be delighted to to chat about that. But yeah, that's the best way to connect with us. And emotionize ai, of course, has our own linkedin page as well, if you'd like to have a look Fantastic.

Speaker 1: 23:37

What I'll do is I'll add in all the links in the body of the webinars and the podcasts and everything else. Marie, thank you so much. Indeed. I think this is something that all of us could do with a little bit more of that little bit of support when it comes to emotional intelligence. At times and that's no criticism of ourselves, as you say, everybody goes through different things every day but your customers coming to you not knowing what you've actually been through a feeling, they've their own emotions and you want them to come away with the best possible experience and any AI that we can apply to that to augment talented, we hope, individuals with new and emotional skills, all the better.

Speaker 1: 24:15

Thank you so much indeed my pleasure, kieran.

Speaker 2: 24:18

Thank you very much.