Listen: Next in Tech | Episode 137: Impressions of AI in Society

View Full Transcript

Eric Hanselman

Welcome to Next in Tech, an S&P Global Market Intelligence podcast with a world of emerging tech lives. I'm your host, Eric Hanselman, Chief Analyst for Technology, Media and Telecom at S&P Global Market Intelligence. And today, we're going to be looking at some of the attitudes in society around AI, really perceptions and expectations about what artificial intelligence is going to be doing for us, especially in light of all of the furor around generative AI.

And joining me to go over some interesting data, to talk about some of these trends, our returning guests, Sheryl Kingstone and Alex Johnston. Welcome back to the podcast to you both.

Sheryl Kingstone

Thank you, Eric. Yes, it's a great topic.

Alex Johnston

Yes, look forward to talking.

Eric Hanselman

Well, I thought the interesting thing about this is that we've got some longitudinal data that goes back awhile because we've been asking respondents, and this is part of our Voice of the Connected User Landscape study, about the impacts of AI for quite a while. But yet now, we're into, I think, I'd say, a slightly different set of expectations, a new day and age, as it were, that brings in generative AI, but we've been seeing a set of interesting changes in terms of expectations, not only from the broad sample but also some interesting variations between the way different generations are both expecting to use AI, put it to work and how they feel about it. But that's something you've been digging into, some of this data, yes, Sheryl?

Sheryl Kingstone

Yes, absolutely. I mean, when we last discussed it in that it was right as it pops. So as you know, Eric and Alex, longitudinal data is boring until something actually happens. And so as we continue to field on general AI, everyone had already accepted that fact. They were starting to understand it, they were starting to use it. It was getting embedded in a lot of things. And then boom! ChatGPT, OpenAI, generative AI changed everything. And so that's when concerns and negativity skyrocketed through the roof, right?

Then we decided, okay, let's add it in another quarter, because we were doing this twice a year. So this time, we got quarter-to-quarter type, you would say. And honestly, overall, negativity is still high. There are some nuances, only looking at Generation Z in the top box, their attitudes towards changing to a little bit more positive that doesn't mean they're mostly positive, all of them. And like we'll go into a little bit more of the details. But as you said, there is one area there. The other interesting aspect though is the awareness factor, right? So there are a lot less don't knows. But overall, as Alex has done with some of the research when we added it up, the negativity is still there. So Alex, you saw when you were correlating the trends.

Alex Johnston

Yes. Sure. I think I'm just going to -- almost cementing what you just said, Sheryl, in the sense that from Q4 2022 to Q2 2023, which was a massive step change. And we're effectively tracking 2 things the level of impact that people associate around their personal lives, career, society but also how positively or negatively they receive that change or they perceive that change. And we've not seen the same sort of attitude nor step change we did previously, we've almost seen a hardening of that trend.

So as Sheryl said, there's a growing awareness of the fact that AI will have impact. Notably, less respondents are not aware of AI or machine learning. But actually, the sort of distributions of organization is quite similar, or I should say, individuals, it's quite similar in the fact that positive or negative considerations largely reflect our results earlier in the year.

Eric Hanselman

Interesting.

Sheryl Kingstone

Yes, absolutely. And when you're talking about the differences across society, career and your personal live, the biggest change and the biggest difference really was on society. That was the mostly negative impact. A lot of times, people don't necessarily see how it's going to impact them. They don't necessarily see how it's impacting their career.

But they can see in the news that society is really making that change. And then overall, it has pretty much leveled out. But there is this, I would say, 4- to 5-point significant impact change when we're talking about those 18 to 24 year olds, the Gen Zs because we don't really go younger than that. And then the rest of it is kind of normalized middle of the road, right, somewhat positive or mostly negative, but that's really where it has changed out.

So the difference between mostly positive on their career for Gen Z was up by about 5 points. Same with mostly positive on the personal life. They're using it. Everyone else pretty much stayed the same, just hardened with the road as it towed and the don't knows became more don't knows. But again, even society, they were much more positive probably about 5 points. But that's only in the top box. Everything else pretty much normalized out and stayed pretty negative.

Eric Hanselman

Well, that's interesting in that when you think about the fact that the generative AI message really happened in November of last year where things really took off. There was a lot of very negative reporting about societies doomed, is the next Armageddon going to be kicked off by generative AI. There were some pretty downbeat messages that were happening at that point. And yet, we still seem to be in a situation where so far, there seems to be relatively positive impact.

Sheryl Kingstone

Only in that one area, we're still mostly negative across the board though. Let's just make sure that we understand when we talk about longitudinal data, there was a pop. And now there's, as Alex had said, a hardening and there's one little positivity nature of it, and that's the Gen Z. And let me tell you, we could refill next quarter and they could change their mind. So...

Eric Hanselman

So the rest of us really still think that Skynet is coming and that we are all doomed. Okay, good. All right. I'm glad it wasn't too far off there.

Alex Johnston

Well, just a quick sort of guide against that, Eric, we still do see more optimists than pessimists in terms of impact on career and personal lives.

Sheryl Kingstone

Yes, true.

Alex Johnston

And on society, it's neutral. But Sheryl's point is there's massive negative turn. We saw previously quite a serious net positive, particularly, on certain demographic groups, for example, higher educated respondents and it's trended quite negative since the introduction of generative AI. But there's still marginally more optimists than pessimists on AI. But it's far closer than it used to be, say, 2 years ago.

Eric Hanselman

So well -- and actually, I guess, maybe we can pause for a second because this is a study data that goes how far back now?

Sheryl Kingstone

Couple of years, like back into 2019 is when we polled it.

Eric Hanselman

Okay. And in that realm and that day and age, we were talking about AI in ways that we're really -- we keep talking about traditional AI or legacy AI, but it was something that really was much less pervasive...

Sheryl Kingstone

Correct.

Eric Hanselman

In terms of a societal understanding about where that was. And you were seeing that in terms of an understanding of the don't know category of what is AI all about kind of piece. So you saw an uptick of that. So there is now greater awareness with this explosion of all the messaging around generative AI, but those attitudes have not changed dramatically with all of this extra discussion that's going on around AI and the pervasiveness of generative AI in it seems like just about everything we do.

Alex Johnston

And the pervasiveness point is quite interesting. We found that only around half the population doesn't use generative AI tools and have no plans to. And when you break that out, around different generations as Sheryl said Gen Z, in particular, is an area we've seen quite a lot of usage and less than 1/4 of Gen Z haven't used some generative AI tool. And only around 1/3 of millennials haven't. With the older generations, it's around 82% of those born before 1945, they haven't been using the tools. But the pervasiveness of the technology is, perhaps, even greater than we'd realized initially.

Eric Hanselman

Oh, interesting. Well, I guess we start wrestling with some of these kind of questions about what are levels of perception, what are levels of understanding. They're using it, but I thought one of the other things that was interesting about this was also an indication of frequency of use.

Sheryl Kingstone

Right.

Eric Hanselman

And in fact, how often they happen to be using it, not only just have they ever...

Sheryl Kingstone

The majority of users when we look at the ones that are actually using it, right, so it drops it down tremendously when you back some of those out, down to around 80% daily, right, at least once a week up to like 33% on total, a few times a month. It was only around 12% that said very infrequently or almost never, and that was dominated, as Alex said, by the baby boomer and Greatest Generation from there.

Men tend to be a little bit more of the daily users than women. And then when we take a look at things like income, honestly, higher income users are actually using it. They're testing it out. The ones that are employed or self-employed are using it more. So we're starting to understand and use these in our daily lives. And some of the top use cases, though, go across our work and personal lives, general information search, right?

So while we were really looking at that potential use cases, whether it was for the variety of personal use cases, whether it's for work or professional, whether it's for school work because we were talking to some students here or just personal use, it was pretty much either a wide variety of personal uses. No one said just only professional, around 18%, very small. The top, as you expected, was around general information search, writing assistance, games and entertainment, shopping assistant, objective advice to solve a problem. We can talk about whether it was truly objective in a minute. Creating images, right?

So what's interesting is when you look at all these use cases we threw at these respondents, what was lower was they didn't necessarily know how to use it to guide them within like a software application things that we're spending our daily lives looking at. Where is Microsoft Copilot going? What's Adobe Firefly doing with like embedded photoshop. There's a lot of great productivity there that is not really exposed to the end user yet, and we're still going to start playing around with that. And it will be interesting to see if that picks up over the next couple of quarters.

Eric Hanselman

Well, I wonder, given that we've now got a set of easier-to-consume tools. We've now got apps for phones that will do that. We've now got ways in which people can actually reach out and interact with it directly as opposed to previous generations of AI, you really had the AI aspects embedded into that thing.

You had your Netflix recommenders. And I guess I wonder about the extent to which now that people can actually touch it directly in a much more easy-to-consume fashion now they're dabbling in it. But I think the interesting thing about the data that you pointed out is that they haven't gotten to that next stage of actually using it to -- using the leverage it can provide to be able to use a software application, to be able to help with creation of content.

They're still doing search. They're doing bits and pieces of things, but it sounds like we're still in the early days of the population at large understanding how to do interesting things with it.

Sheryl Kingstone

Well, we know that. So as human beings, we like to transition into something that's familiar. So what's familiar is our Google searches. And so now we're taking that use case that we're very comfortable with and transitioning that into ChatGPT. Alex, help me out here. So until those other use cases become more embedded, they won't understand how to use them.

Alex Johnston

I think something to keep in mind as well is a lot of the integrations and partnerships have only kicked off in the last couple of months. So it'll probably be a while before we see what we're likely to see, which is a conversational interface, being a way that we're engaging with a high degree of software that we use as consumers. But as Sheryl said, I think what's really interesting is you've got these almost 2 polarized communities. You've got around half of the population not engaging with this technology. And then of the percent that are, about 50% of those individuals are very heavy users. They're using it more than once a week, particularly if you are younger individuals, the millennials and Gen Xs, you're just talking about 58%, 55% using it once a more a week.

So those that are using it are using it quite heavily. And actually, whilst a lot of very exciting applications aren't in use, if you look at the wide array of ways that people are using it, so general information search, writing assistance, as Sheryl mentioned, people tend to be using it in multiple ways, which is impressive for such a nascent technology.

Sheryl Kingstone

Yes. Absolutely true. We keep forgetting and bring reality too. It's been only a few months, right?

Eric Hanselman

It just feels like it's been that much longer...

Sheryl Kingstone

Students are using it. They are helping with -- 32% are helping them with their school work.

Eric Hanselman

This we know in a number of fronts.

Sheryl Kingstone

Wait, wait, wait, 13% virtual companionship. Now these were future use cases too. We asked what do you do today or would be interested in, in the future?

Alex Johnston

Something else which stood out as well, obviously, multimedia or multimodal generation is a big talking point in terms of how the trend is evolving. So a lot of them, most popular application areas are a text generation. But actually, one of the great parts of our survey, I looked at areas like image generation, video generation, music creation. And we do see interest in all of those areas, perhaps unsurprisingly, image generation outstripping the other nontext areas being a slightly more mature market. But it's interesting that a lot of the, say, create music, for example, is a very, very immature market. So it's interesting that people do have quite a lot of interest in those sorts of areas that really stood out on the data.

Sheryl Kingstone

Yes, absolutely. I mean just below the general help with your copilots and Adobe Firefly, creating images, you could say Adobe Firefly is actually embedded in that, and that was up towards the top. And I've used it, it is fabulous when you're trying to do some Generative Fill and some guidance and summary angle of it. We've always had some of that. It just becomes a little bit more intuitive. So we will see how it really matures.

Alex Johnston

We also collected some fairly interesting data on concerns around the implications of generative AI, which perhaps add some flesh, I suppose, some of those headline stats we're talking about before.

Sheryl Kingstone

Yes. Good point.

Alex Johnston

But unsurprisingly, I did find it quite interesting that was not a major concern overall. Gen Z were the most likely to worry about AI becoming conscious, which I thought was, perhaps, a reflection on the sort of media that people absorb at the moment, where scams and fraud were far larger concern for older generations as well as risk to data privacy.

Sheryl Kingstone

Yes. So put that into context, we gave them a list of potential concerns. So they didn't come up with it on their own and could become conscious or capable of feeling what's the bottom of the road. But Gen Z was actually surprisingly the highest concern there for their generation per se, the biggest concern for that generation was risk to their data privacy, right?

So that beat out scams and fraud as compared to the Greatest Generation is always concerned about scams and fraud because they get taken advantage of. So there's also experience with that. Alex, the other thing is when we look past the ones that everyone understands, the scams, the risk to data privacy, the misuse with ill intention, right after that is the disinformation, information incorrect use, deep fakes and that's something to be concerned about. You've actually discovered some of that yourself.

Alex Johnston

We have looked in quite a lot of depth. Obviously, there's some regulation that's being particularly set up to address that or attempting to address that. There's a lot of conversations in U.S. in particular about the role MI play and political campaigns, for example, there's a lot of concern there. But I think one of the areas we're seeing companies look at is things like water marks, for example, to make it quite clear when something is regenerated rather than a natural thing that's actually had a photograph taken of it, for example. So there is a lot of people looking at this area.

Eric Hanselman

And that's something where if we think about what watermarking would do, it would mean that you could identify that an image was generative AI generated and understand what its history was, be able to actually figure out providence in some form or fashion.

Alex Johnston

Yes. One of the other concerns that was quite interesting was a relatively small proportion of people were concerned that AI would replace them at their job. But I think that's a reflection of the fact, that's people have experimented a bit more with these tools. The way the maybe this technology might play an augmenting role in the fact by helping some of these areas that Sheryl mentioned already around sort of search, for example, but it's unlikely to be in a position that it will actually replace many roles. I mean at 22% of people being concerned about that, it's far smaller than say the amount of media attention that's been provided to this topic, for example.

Eric Hanselman

It raises an interesting point, which is, is this something where we've got the situation where the people who work with it realize what its limitations are and understand what -- how far I guess it can go in some form or fashion. I mean that's something that on the analyst team we see all the time, something where, hey, for generalized search for things that have got long histories, where there's rich sets of data, it works reasonably well.

But yet, if we're looking at areas that are maybe a little more forward-looking, places in which there isn't as much history, performance isn't quite there. And I guess those are things that maybe this body of respondents has actually started to see and maybe that level of concern or the lack of a higher level of concern is due to the fact that they're actually starting to understand what those limitations are.

Sheryl Kingstone

Potentially. Yes, absolutely. One of the things I do want to point out as we're here on the whole concern issue, getting back to the generational issue, when you go from the young ones to combining even baby boomers and Greatest Generation, you're doubling. So Gen Z was around -- like most of them were fairly low. They were fragmented. 40% concern on scams and roughly risk to privacy. You're talking 78%, 61%, a lot of them were really very, very deeply concerned, right? So it leads to the correlation between some of these concerns and then going back to the original first conversation we had on the impact, right? So we are definitely seeing a difference in a [ bipolar nature ] between the old ones and the young ones.

Eric Hanselman

So strong generational shifts in terms of where those concerns lie and presumably as a result, what their receptiveness is to adapting and working with the technology.

Sheryl Kingstone

And surprisingly, Gen X, who squashed in the middle is in the middle of the two.

Eric Hanselman

The middle of the road is the middle of the road.

Sheryl Kingstone

Yes.

Eric Hanselman

So I guess just to take a step back, in terms of looking at this data, so this is ongoing data that we're going to be continuing to refresh as we go forward. In terms of how people should leverage it, what do you think are the takeaways and what should enterprises broadly be thinking about this? It sounds like it's an indication that there is some weariness, depending generationally and how you look at this. There are aspects of that there's value seen in it. But what are some of the takeaways from the business side that people should be thinking about?

Alex Johnston

My big concern is this is going to start generating some change management issues. I think it's not really been adequately spoken about, but I think there'll start to be concerns around user rejection, for example, if people are concerned around their career, for example, which is something we are seeing in data. And most people might not be scared about their whole jobs being replaced. That might become more of a concern when we start to think about generative AI no longer generating a single template of content, for example, we start to think about its role in sort of multistep workflow automation, for example, those fears might start to become even more pronounced. So I'd find particularly with older staff, which, as our data tells us, are more likely to hold concerns around this technology. I think organizations got to be very careful as to how they couch their investments internally.

Eric Hanselman

So this is something where messaging needs to be clear about what it's doing, what it's not doing and what this transition looks like?

Sheryl Kingstone

Yes. But the vendor ecosystem, the tech ecosystem is doing this, at least in my world, because we have to when we're dealing with customer data, putting in some platform changes for data privacy, they're making sure that the tools have some sort of hallucination controls, bias controls, customer data redacting.

So there's a lot of ways that the ecosystem is coming together to make sure that we're protecting their businesses and their customers for one. For two, they're also embedding, and this is why I'm really going to start to watch looking forward what the vendors are doing to be much more prescriptive with the guidance because there's so much opportunity as we become much more data-driven to use a lot of the technology for positive outcomes. Call wrap-up reports in customer service and support industries. That's huge.

Because writing assistance, we're really trying to get things done. It really drives efficiency and cost reductions there. Using applications across the board from new Gen Z finally using some of these tools that are forced on them, especially with like a Microsoft Office that they might not have been using, they were using more Google tools in the past, Adobe Photoshop, all of those tools with the creatives and really embedding more of that workflow and guidance next steps is a huge opportunity.

Now if we take a look at the SaaS industry and using all the data to potentially improve and do some benchmarking on how we're using the applications more effectively. So there's a lot of negatives, but there's also a lot of positives if we are understanding and open to how it can be much more guidance and help them in their job and their professional work life.

Eric Hanselman

To Alex's point, there has to be understanding and communication that gets brought along with this to ensure that people know what they're getting, know what this augmentation looks like, deal with concerns about job loss and make this part of that larger whole?

Sheryl Kingstone

Yes.

Alex Johnston

I think it's also quite a unique technology area, and these adoption stats really back that up. And often, I'll talk to organizations who are saying what should our first steps be with generative AI, in reality, your staff are already using these tools. We have to have a more -- it isn't like, say, blockchain, for example where it's unlikely that, that many people besides maybe some experimentation with cryptocurrency, people are going away and doing these things.

It almost feels like we have to have a reactive governance structure to respond to the fact that people are going to be using these tools and if they're not sanctioned variants, they'll be using the public-facing alternatives. So I think it's quite a unique emerging technology area in that sense.

Sheryl Kingstone

Yes. Except for some of us, Eric, if we go back 2 decades in the entry of SaaS and how that started with grassroots because you no longer had to go to IT to buy anything. So it turned out and started the whole trend of the consumerization of the enterprise back in the early 2000s.

Eric Hanselman

So we -- once more, we're back into this technology cycle of early adoption of specific new use cases in enterprise use, we then have to figure out how to manage.

Sheryl Kingstone

Yes.

Eric Hanselman

And getting back into the same kind of things. But Alex, to your point, they need to come at this with an expectation that their employees are already using this and now how do you transition that into sanctioned use, how do you educate about intelligent use, secure and effective usage of the technology and how they actually integrate it.

Sheryl Kingstone

Yes.

Eric Hanselman

Well, a lot to cover, and I guess we'll see where this goes. I think as we've all been discussing, this is such early days in this, but so interesting to see where this is headed and a lot of these adaptations about expectations, but certainly something that we'll be discussing on future episodes. So thanks to you both.

Sheryl Kingstone

Thanks, Eric. I really love coming to your podcast and discussing all this unique data that we have.

Alex Johnston

Thanks very much for having us, Eric.

Eric Hanselman

And that is it for this episode of Next in Tech. Thanks to our audience for staying with us. And thanks to our production team, including Caroline Wright and Kaitlin Buckley on the Marketing and Events teams and our agency partner, the One Nine Nine.

I hope you'll join us for our next episode where we're going to be looking at in-depth discussions about technology and some of the things we get alongside some of our research. I hope you'll join us then because there is always something Next in Tech.