The Future of AI with IDC's Hayley Sutherland

ABOUT THIS EPISODE

Speech-to-text has improved a lot. It doesn’t mean that AI is the modern-day Skynet, seeking to steal all of our information or take over our world.

Special Guest Host, Patrick Ehlen, interviews Hayley Sutherland about what’s really going on in the world of conversational AI. Hayley is the Senior Research Analyst, Conversational AI, at IDC.

In this episode we:

-Reflect on the past 12 months of AI advancement

-Break down AI’s response to highly contextualized situations

-Forecast AI’s greatest milestones for the next five years

Keep connected with Conversations That Matter at Apple Podcasts, Spotify, or www.uniphore.com.

Want to be a guest on Conversations that Matter? Email podcast@uniphore.com.

When we speak to machine as wewould to each other, they're actually starting to understand what we mean and thecontext behind that. Welcome to the conversations that matter podcast from uniform, thepodcast that dives into real conversations and that are happening in contact centers around theworld. Here you'll experience exciting interviews with wellknown thought leaders. Here, compellingstories from industry experts gained fresh insights on context and our best practices. Inmore so, grab a beverage in tune in as we get real with conversationsthat matter. Hi, I'm Patrick Ellen, vp of AI uniform, and I'mhere with Haley Sutherland, who is senior research analysts at IDC, andwe're going to talk a little bit about Ai and what we've seen over thepast year and where we think it's going. And Haley's a great person to havehere to have a little chat about it because she's IDC's expert on Ai. Welcome, Hailey, Hi Patrick, it's great to be here with youtoday, virtually so, as Patrick Mentions, I'm a senior research analyst here atIDC. We are a technology research...

...firm, which means I get tohave lots of interesting conversations with smart people about artificial intelligence and I'm very excitedto bring some of those conversations here today to our discussion. Patrick great oneof the things that I've noticed over the past year is this catch phrase thatseems to be going around where people say I'm shocked but I'm not surprised.You see people saying that I'm news all the time, usually in reference topolitics in the day's news. But I'm wondering, you know, if wecould sort of think about ai over the past year and are there things whichhave happened that have made us shocked but not surprised, or only in thesurprised category, or perhaps in the shock down surprised category? I'm interested toknow what you think. Yeah, so, you know, I think that's agreat question. I am a huge story nerd and you know, Ithink any good story you get to the end and what you think of thiskind of surprising yet inevitable. You know you didn't expect it, but onceit happens you go off of course, and there have been a few thingslike that this year for me that I...

...think we're really interesting. You know, I think one that a lot of folks are may be aware of ourall the advances we've seen in natural language processing and open source and and thosehave been just huge, hugely to the last couple of years. Think aboutthings like burked, you know, transformers, all of the many burnt derivatives thatfolks have come up with. You know seeing places like hugging taste usingtransformer technology to quickly implement ML models, machine learning models, and so whatthis all means is that conversational AI systems are really starting to understand context andintent in natural language. You know, when we speak to machine has aswe would to each other, they're actually starting to understand what we mean andthe context behind that. And the other thing that we're seeing, you know, as part of that, especially with all of the events, is inopen source, the advances in machine learning and deep learning that are making trainingless of a burden in terms of the time it takes, the resources,in terms of people, the compute power...

...of the data. We're really gettingto more accessible technology. There's higher accuracy with less data, there's less ofa burden on humans to manually create ontologies and taxonomies and it's really been incredibleto see. At I see we've done a number of surveys over the last, you know, ten months or so to really understand, you know,how businesses are changing, how their responding to covid nineteen. And, youknow, a big one that we asked about was in terms of contact lessexperience. Is what our organizations planning, and I was pleasantly surprised to seethat the top planned investment to support contactless experiences is actually voice based interfaces.So that was, you know, again, really cool to see. But itmakes sense when you think about it, you know, in terms of what'sgoing on in the world. So yeah, that's very interesting statistic.We did a survey here at uniform over the past year just sort of askingpeople how they feel about a in general and and talking to, for example, a live agent versus a an AI...

...based agent, and one of thethings in our survey that we found was at seventy six percent of consumers whoare out there still prefer to speak to a live agent rather than, youknow, talking to an IVR system or even a fairly smart ai system,and I do think that a lot of that is is just because ai isnot yet really taking into account sort of the broader context that is happening notonly within a conversation but within a situation, within people's lives and and all thatsort of thing. I think another statistic from that was that only aboutthirty two percent of people said that they felt comfortable or in some comfortable,using AI for for customer service, and I really think that it is abog part of that is just the fact that ai has yet to be ableto kind of, you know, look at you on zoom or on youryou know video or sort of your situation that you're in right now and takeinto account a lot of the things that...

...are going on in a situation.What do you think about that? Yeah, I think that's true and I thinkthat's that. I think there are a couple of pieces to that.So, you know, I think part of it is, you know,trying to get that better understanding of context and intent through natural language processing,you know, understanding that when someone says, well, that was hardly helpful,they really mean that wasn't helpful at all, and it was starting toget better at that, but I think you you know, to your pointthat trust in aizes ability to do that is going to lag a little waysbehind. That lines up with the conversations that I have, you know,with with bier organizations, with any user organizations, you know, even thethe ones that are using ai and that are fairly mature, and it willsay, I love to ask them what challenges they're having and I always get, yeah, can you talk to my colleagues and let them know, youknow, a it can do. You know, let it know what theycan do, that it can, you know, start to understand these thingsbetter that it is, you speech to text has gotten a lot better,those kinds of things, you know,...

...but also help them understand what itcan't do, right, that it's not Skyne it and it's not going tocome and, you know, seal all their information and take over the world, and I think that's really impelling that we're still kind of at that phase. But another piece of it is, you know, you mentioned kind ofthat, you know, that human aspect, right, being able to understand whatthe person you're talking to is thinking and feeling, you know, kindof in both ends, to feel like they're listening to you, hearing tohearing you, responding to you and to your emotions and to your context,and I think it's going to be really interesting to see. You know,one of the areas that I look at is emotion AI, or effect ofcomputing, which is, you know, teaching ai to to understand human emotionsso that maybe some of these conversational systems can actually respond a little more empatheticallyto a frustrated customer or, you know, to someone who's maybe taking out aninsurance claim for something really, you know, emotionally painful that's happened.So I think that's that's a really interesting question, especially as we see moreand more digital assistance proliferating, you know,...

...both in the consumer world and startingto move into the enterprise. You know, they're doing really well withthings like task automation, but how can we get them to be more empathetic, more human like? And it's been very cool to start to see someof the advances in emotion AI and natural language processing that are starting to makethis a little more possible and starting to make those things, you know,kind of emerge. Now for a short break, let's learn more about uniform. Uniform is a global leader and conversational service automation companies vision is to disruptan outdated customer service model by Ridgeon the gap between human and machine using voiceai and automation so that every voice on every call is truly heard. Formore information about uniform, go to www dot unifordcom, email podcast at uniforcomor tweet at uniform. And now back to the podcast. Yeah, Iabsolutely agree and I think that's that's a...

...huge step towards the kind of likebeing able to pull in more context into the conversation that you're having with ana I than, you know, than what we had before. But Ialso think that that's that's only the first step. I mean having a machinethat can understand emotion from your facial expressions and also from your voice the theway that humans do is definitely really cool and you know, sounds like thefuture, but I think it's actually the future that's that's here today. Butthe real future, I think, is also in and I think you knowwhere we hope to go is being able to integrate more in terms of boththe speech and sort of the video and the context that you're getting from thevideo at the same time, not just somebody's facial expressions but, as Isaid, the entire context, you know. So I don't know, you know, I'm that's sort of my vision of the future. There is reallyjust ai that is not just doing speech, not just doing vision, but istaking all of that stuff and putting it together and kind of understanding thehuman society that we live in in a...

...much better way than it does rightnow. What are your prognostications for ever, let's say, the next year orfive years or or whatever? Where do you think we're really going togo? So yeah, and that's great timing because every year at IDCON gota set of predictions that we call our IDC future escapes, where we look, you know, three to five years ahead and you know, anywhere fromfrom one year out to five years out to try to understand, you know, what's going to happen in the near term, in term and long term. That this is is should be planning for, and we have a coupleof interesting predictions there that I can share. You know. I think one isis really about the pervasiveness a AI. You know, despite the challenges ofmoving things from pilot or proof of concept into production, we are seeingmore and more organization starting to be able to really solve that problem. We'reseeing more models than ever moving into production and one prediction we have actually isyou know again really about how AI is everywhere. We predicted by tw thousandand twenty two, at least sixty five...

...percent of the G two thousand willbe using AI, such as natural language processing, machine learning and even deeplearning. Business. Why? To An abl we think sixty percent of usecases across various functional areas, and that's from, you know, things likeprocurement to HR, to customer service, you know, all these, allthese things that I think, you know, ai can can really help to,you know again, automate tasks, answer questions, you know, andyou know, just make us smarter, help to Augment People's capabilities, youknow, rather than replacing them. So I think, you know, we'regoing to see more and more humans and machines working together and I think that'sgoing to start feeling a lot more normal. I think we're going to start seeingsome of that fear of being replaced hopefully disappear and, you know,and change into hopefully, excitement about, oh, I don't have to dothat in manual task any yeah. Yeah, I think that's another important point.Is like, you know, the more people feel like it's not aboutgetting replaced but it's about having something that you know, is going to doa lot of your work for you or make it a lot easier for youthan the easier it will be for people...

...to adopt it. Yeah, yeah, absolutely. I think another interesting thing if we look further out, youknow, that's where things really start to get interesting, because research is comingout all the time to AI. Research world is just no constantly surprising andamazing me with new models, new methods, new ways of doing things. So, you know, one of my predictions, is that this is justa personal prediction, not a future skate prediction, is that there will bethings happening in AI and five years that we couldn't have predicted right now,which I think is really cool to think about, you know. But interms of officiality se future skate predictions, we predicted by two thousand and twentysix twenty percent at Ai will be closer to artificial general intelligence. So thinkof that as kind of, you know, true reasoning in the way that wethink of it in terms of how humans do it, you know,being able to make kind of logical leaps and reasoning, being able to piecetogether information from multiple sources to arrive at one complete answer, things like that. You know, I think we're still...

...quite a ways and even in fiveyears will still be quite a ways from, you know, kind of an allthinking, all knowing, you know, context switching artificial intelligence. But wedo think that, you know, in five years we are going tosee, you know, situationally. The AI is going to get a lotbetter about that and I think going to be really have an impact on conversationalai, especially in terms of, you know, the question answering, taskautomation and again that that kind of worker augmentation. So so really cool thingscoming, I think, and you know again, as as we see moreand morally digital assistance. I'm really interested to see. How are they goingto work together? How are they going to talk to each other? Arewe going to know how to write, you know, route the right requestto the right digital assistant? I think it's going to be very interesting.Okay, wait, that sounds very excited. I'm looking forward to that future.Well, Haley, I want to thank you for taking the time tocome talk to us in this has been a super interesting conversation. Well,thanks for bringing me into it, Patrick.

I always love talking to you folksat uniforms. Always a good conversation and let's keep it going. I'mreally excited to see what the next, you know, three to five yearsbring. Yeah, yeah, let's let's see what two thousand and twenty oneis like. Do you have been listening to the conversations that matter? PODCASTby uniform. Make sure you subscribe to our podcast on your favorite podcast playerand rate in review to enable us to create relevant and valuable content for yourbusiness. If you'd like to learn more about conversational service, automation is theuniformcom have a great day.

In-Stream Audio Search

NEW

Search across all episodes within this podcast

Episodes (26)