This post was originally published on this site.
Episode Transcript:
Marcus Johnson (00:00):
Struggling with out-of-stocks, phantom inventory, or lack of shelf data? Trax’ signal-based merchandising is designed to give real-time visibility into what’s happening in stores and on shelves so you can take action when and where it matters most. Interested? Head to Trax, T-R-A-X, retail, traxretail.com, and click the get started button in the top right-hand corner.
(00:25):
Hey gang, it’s Monday, March 31st. Gadjo, Henry, and listeners, welcome to the Behind the Numbers Show, an eMarketer Video podcast made possible by Trax. I’m Marcus. Today we’ll be discussing using AI at work. For the conversation I’m joined by two folks. Let’s meet them. We start with our senior analyst covering AI and all things technology. Based in New York, here’s Gadjo Sevilla.
Gadjo Sevilla (00:46):
Hey Marcus. This is exciting.
Marcus Johnson (00:48):
Hello, sir. Thanks for being here. And we also have our SVP of Media Content and Strategy. Living in the land of the lighthouse, it’s Henry Powderley.
Henry Powderly (00:56):
Hey Marcus.
Marcus Johnson (00:57):
Hey fella. Today’s fact, why do we blow out candles on top of birthday cakes?
(01:06):
Here’s why. So according to Alysa Levene’s book Cake: A Slice Of History, great, great book, historians credit the origin of birthday cakes with the ancient Greeks, explaining that it was a way to honor the birthday of Artemis as the goddess of the Moon and they would bake moon-shaped cakes decorated with candles to make them glow like the moon. And the Greeks believed the candle smoke would rise to the heavens where Artemis, the daughter of Zeus, lived since offerings to the gods were a common custom in ancient Greece.
(01:39):
The reason we put the same number of candles on the cake as the person’s age is because of our German friends. German Kinderfest, or children’s festival, basically, kids’ birthday, dates back to the Middle Ages and the cake was topped with lighted candles which were presented to the kid on the morning of their birthday. They used to have the candles just burn throughout the day until the cake was eaten after the evening meal. Fire hazard. But now it’s obviously it’s changed. They blow them out straight away. According to The Extraordinary Origin of Everyday Things, that’s a book by Charles Panati, the number of candles on the Kinderfest cake equaled the kid’s age plus an extra one representing the light of life. Which makes blowing them out a little bit dark.
Henry Powderly (02:26):
I like one for good luck better.
Marcus Johnson (02:27):
Pun intended. Yeah, exactly.
Gadjo Sevilla (02:29):
You always need a spare, right? I guess. I don’t know.
Marcus Johnson (02:32):
Spare candle?
Gadjo Sevilla (02:33):
Yeah.
Marcus Johnson (02:34):
Yeah. But what’s the year? Like at five you stop putting on the same amount as the person’s age? When does that stop? Teenager?
Henry Powderly (02:44):
Depends on the size of the cake, I guess.
Marcus Johnson (02:46):
Yeah. Or your parents. My mum was like, “We’re done here. Okay?” Six, she cut me off. That’s a joke. Kidding. Love you, Mum. Anyway, today’s real topic, using AI at work, part one. Employees anxiety levels and how the technology is impacting jobs.
(03:06):
All right, for this episode, it’s the first episode of a two-part episode. Second episode will be on Friday, April 4th. Today we’re talking about how the technology is affecting people, affecting the workers, and how it’s impacting jobs as well. We’ll start with the workers piece of this. So it appears as though anxiety around AI in the workplace is getting worse. There’s a recent Pew research study noting that more Americans are worried, 52%, than hopeful, 36%, about the future use of AI in the workplace, current and future use. The gents are going to take differing arguments for us. So these opinions don’t necessarily reflect their personal views on the topic. But Henry has volunteered, I made him, to present the argument that folks will become less worried over time. So it seems like they’re becoming more worried, but Henry, what’s the argument that actually will peak and end up going in the opposite direction?
Henry Powderly (04:07):
Sure. Thanks for telling everybody that you made me do this. As a caveat, I do think that there certainly aren’t types of jobs, especially ones that are more automated, but I do think there’s reason to worry in the future, but most of the jobs that we’re talking about, especially in the eMarketer audience are from the knowledge work universe, and I do think that there’s reason to believe that the anxiety will lessen over time. And I think it really will boil down to two things. The first being people actually finding help in their job using GenAI tools. The more people use this stuff, the more they find efficiency, the more they are able to offload those low-hanging fruit tasks that have made the onerous part of the workday, the more that they find themselves have a thinking partner by using GenAI to validate their strategies, to check their copy. There’s so many different examples that we could talk about where AI can help someone in the workplace. But it’s only when somebody actually feels that relief and feels like they’ve gotten that help when they’re going to start feeling less anxious about it.
(05:13):
And then the second side is on the organizational side. When people start to see that help in their work, are they met with more resources or more opportunities to grow in the organization? Do they feel like they’re getting enough training that’s enabling them to find that kind of relief and creativity that could be found through using GenAI? I think it really will depend on organizations pushing in that direction. And then how they react to people finding more efficiency in their work. Is it going to be a reaction that’s downsizing or is it a reaction where investment is made so that we can use the time that people have bought back by employing generative AI to perhaps, I don’t know, explore new products, level up the parts that they have. All the creativity that can come out by using the time that they’ve saved.
Marcus Johnson (06:05):
That appears to be a huge concern, doesn’t it? That if I use AI and I make my job more efficient, then I’ll need less time to do it, which means that maybe they could get rid of me and have someone else do what’s left of my job. There was some research on this actually. Few workers think AI use in the workplace will improve their job prospects in the long run, 32% of people expected to have fewer job opportunities, 6% expected to have more. And interestingly that was both AI users and non-AI users both appearing to be pessimistic. If you looked at non-AI users, it was a six to one ratio in terms of people who think that they’ll have fewer job prospects versus more. And if you look to AI users, it was three to one. So it’s still concerning for both groups.
(06:51):
You also mentioned that people maybe haven’t realized the true potential of it yet. It’s not enough to just use it, is it? There has to be that light bulb moment where like, “Oh, this is how it’s going to help,” which is a lot further down the road. There’s still a lot of people who aren’t using this technology even and people maybe not realizing it’s true potential because they just haven’t used it. Pew study, 61% of people aren’t using it much if at all, and a further 17% haven’t heard of workplace AI use. So this is still very new and we’ve just got to get more people, I guess, using the thing before they can be more comfortable with the thing before they can realize how the thing can help them.
Henry Powderly (07:30):
For sure. And then how are companies going to help them find the time to be exploring these techs? Think that’s a real big part of it.
Marcus Johnson (07:36):
Yeah. So that leads me to another point here, which is the training piece. Folks aren’t, it seems, being trained on its benefits. Of all workers, this Pew study, 12% said they’d taken a class in the past year related to AI use, just 12%. And there was a one Randstad study showing that about 35% of employees, so more, had received AI training in the past year. However, that’s 35% trained on it, 75% of companies adopting it. So there’s a big chasm between people using it, people really understanding the benefits of it, and having the time, I guess, being taught and being allocated time, the time to really understand how it could impact people’s jobs.
Gadjo Sevilla (08:24):
I think that’s really the big issue because there’s the initiative to adopt AI, but there’s also the implementation piece. And part of that is not just subscribing to chatbots, it’s also training. And some companies are falling short on the training aspect while demanding that employees are up on the latest technology. Less than 40% of the workforce say they have access to GenAI tools and less than 60% of those employees engage with it daily. So I think access is a huge problem to solve there.
Marcus Johnson (09:04):
Yeah.
Gadjo Sevilla (09:05):
Especially if you consider there are discrepancies. Because according to Deloitte, 29% of AI-skilled workers are women. That’s a small number considering there are limitations to what is available to them.
Marcus Johnson (09:24):
Yeah. One-third of workers paying out of pocket for GenAI tools because their employer doesn’t provide the ones they want. Half of knowledge workers, basically desk and computer people, using personal AI tools at work. That’s from another survey. That’s from company Software AG. And half of executives saying employees currently are left on their own to figure out GenAI. That’s from a GenAI platform writer.
Gadjo Sevilla (09:48):
There’s a huge security issue with that though when people start bringing in their own AI tools and subscriptions because if they’re dealing with company IP, company data, I mean what’s the next step? Where are they putting that information? Where does that reside?
Marcus Johnson (10:06):
Yeah.
Gadjo Sevilla (10:06):
We know AI is constantly training to different degrees and putting very critical information out there might improve their jobs, but then it also opens up a lot of possibilities that are less than great for the company at large.
Marcus Johnson (10:26):
So Gadjo, what’s the argument that people will become more worried?
Gadjo Sevilla (10:30):
I think the argument, aside from maybe the lack of training and resources, is that we’re seeing a point in time where there’s so many AI tools that it’s becoming increasingly more difficult to figure out which ones they need to be looking out for. And so there’s that. And at the same time, people who are using the tools are training the AI to handle their workflows. So do they feel that they’re actually training the AI to replace them to some extent?
Marcus Johnson (11:06):
Yeah.
Gadjo Sevilla (11:08):
What company can give a guarantee that, “No, that’s not the case. You should see this as an assistant, as a tool. It’s not going to replace you.” But if you look at the bottom line for a lot of companies they’re pushing towards automation to sort of speed up and scale certain jobs.
Marcus Johnson (11:29):
Yeah. There are some numbers on that. Quickly, workers noticing their jobs becoming more automated, 16% say at least some of their work is currently done with AI. A further 25% say whilst they’re not using it much now, at least some of their work can be done with AI.
Gadjo Sevilla (11:45):
Yeah. So I mean that you can’t help but consider that they see that and they become anxious about it.
Marcus Johnson (11:52):
Yeah. Yeah.
Gadjo Sevilla (11:53):
Because they see their jobs changing and at the same time they’re seeing the AI get better at doing certain jobs.
Marcus Johnson (12:01):
Yeah. So Henry, I mean before we move on to the next part of this, which is how it’s impacting jobs. Gadjo made a good point there, which is it is hard enough to know what tools to even be paying attention to, especially if you’re not getting that direction from the company, let alone how to use those tools. And you’re someone who is an early adopter of AI, use it, and you understand it more than the average person. How do you pay attention to the right tools?
Henry Powderly (12:26):
I mean a lot of experimentation. I certainly play with a lot of tools personally that I don’t use at work for the reasons that Gadjo expressed. I think security concerns, especially when dealing with company data or company strategies, shouldn’t just be put into random tools without thought. And we do have a good process internally here for vetting tools, but it doesn’t mean that I don’t play with things in my personal time to really explore what’s out there. And then if I find something that’s interesting that could perhaps help my team, I bring it to the folks here and we go through a process and vetting it. We did that with a tool that we’re using for copywriting right now in my team called Spiral and I went through that process.
(13:07):
So I do think that it’s a catch-22, because you want to be exploring the options that are out there right now, and I think standing in the way of that kind of curiosity is going to slow down adoption. But at the same time, companies got to be smart about setting policy that are going to reduce the risk of data being adjusted by these language models. So there’s a balance there.
Marcus Johnson (13:31):
Yeah. Let’s move to the jobs piece of this. Henry, what’s the argument that AI is already having a significant impact on jobs?
Henry Powderly (13:40):
Well, I mean I think we wouldn’t be talking about it if we haven’t seen this kind of impact already. What’s really cool about eMarketer is that we are always getting data about how the market is changing. And I was just looking in our own chart library to see. And there was some interesting stuff there. I mean there was a survey by Nextiva from January that showed that 40% of respondents were using GenAI for writing to customers. That is a big chunk of time. I saw physicians. Physicians, almost a third of physicians, are using right now AI for things like translating or even diagnostic help. And then there was another survey of retail CFOs by Raven Roberts Research from February that showed that 34% are using GenAI for optimizing pricing strategies based on market dynamics. So I think the data is starting to come in to show that there’s a lot of penetration with this tech.
(14:39):
And there’s downsides too. I think there was a story in The Washington Post last week that showed that computer programmer employment dropping to its lowest level since 1980. That’s one of the areas where I think we’re starting to see the effects sooner because of how well these tools are at coding. Just the recent update to Claude already has just given folks like me who don’t really have a good coding background the ability to create things and program things. And you have to imagine for folks who are actually embedded in the programming work how these things are helping them.
Marcus Johnson (15:15):
Yeah. Matteo Wong of The Atlantic wrote a really, really good piece. And he was saying that tech executives have grown blunt about their hopes that AI will become good enough to do a human’s work. In January quoting Mark Zuckerberg saying “2025 will be the year when it becomes possible to build an AI engineering agent that’s as skilled as a good mid-level engineer.” And Anthropic’s CEO Dario Amodei recently said, “AI will be writing 90% of code a few months from now,” with some human specifications, but still, continuing to say, “We will eventually reach the point where AIs can do everything that humans can in every industry.” Obviously he’s someone who is at the forefront of that and believes that and that would be good for him and his company. But still the idea that people are openly talking about, “No, this thing will be able to do this person’s job or this level of a person’s job, ” this year is quite surprising.
(16:16):
It seems as though it’s also being talked about a lot in terms of not what your current job is, but if you’re going to find another job, that is something that people are expecting you to have some kind of a grasp of and have some knowledge around, AI experience starting to outrank job-specific qualifications. Christine Cruzvergara, Chief Education Strategy officer at entry-level job level platform Handshake says employees and HR folks said they’re willing to take chances on otherwise less qualified candidates if they have AI experience. So that becoming even more and more important, more so than having experience in the actual field.
Gadjo Sevilla (16:56):
Going back to the coding part, it’s interesting that at Google, AI is coding 25% of their code right now. That was last year. I think that’s probably going to increase. That goes hand in hand with a lot of the investments they’re making. So they do have a 14% stake in Anthropic right now, and that’s precisely because of Claude’s advanced AI coding capability. So they’re not just drinking the Kool-Aid. They’re actually applying that knowledge, those transformations, into their own business.
Marcus Johnson (17:38):
Yeah.
Gadjo Sevilla (17:40):
And that should be interesting for anyone who’s following how these companies are using AI internally.
Marcus Johnson (17:47):
Yeah,. So what’s the argument, Gadjo that AI isn’t already having a significant impact on jobs? Maybe we’re exaggerating things, maybe we’re getting a bit ahead of ourselves, and it’s a thing, we’ve been talking about it, but really the impact isn’t that seismic yet.
Gadjo Sevilla (18:01):
Yeah. So I took this argument as devil’s advocate, although I do believe the opposite to a greater extent. But that said, I tend to look at it not just that AI isn’t having a significant impact on jobs, but in certain cases it’s having a negative impact, in certain markets.
(18:21):
So in the aspect say of AI image generation and video generation, we’re seeing a lot of AI slop as I call it, which is basically just a lot of random AI generated imagery that’s being pushed as being official or being pushed as prime content.
(18:51):
And I think that’s the danger there. It’s an overuse of AI, but at a super unregulated level where the quality just isn’t there just to sort of stuff pages with content and images. So that could have a negative impact and make consumers and anyone else kind of mistrust AI or mistrust what they’re seeing.
Marcus Johnson (19:23):
Yeah.
Gadjo Sevilla (19:23):
So in most cases, I think that’s the danger, that’s the extreme. And to get back to why it isn’t having a significant impact in some areas or in some business models, it’s possibly because it’s not being strategically rolled out not as a proper tool. And that’s causing the lag in adoption, the mistrust, and maybe it’s the inability to kind of focus on we need this particular tool to help you save hours or improve customer service.
(20:08):
It’s something that a lot of companies are addressing, which I mean to say it’s going to change. For example, once Adobe’s agenetic AI gets adopted at large because they have specific tools that cater to business use cases, just things like understanding the audience that you’re trying to attract, content production, just speeding up the different steps towards that, helping creatives be more creative and be less focused on production and also experimentation and prototyping.
Marcus Johnson (20:55):
Yeah, yeah.
Gadjo Sevilla (20:57):
But all that takes time to adopt. So maybe it’s not that it’s not having a significant impact, but it’s just taking longer.
Marcus Johnson (21:05):
Yeah. I’ll give both sides of this in one answer, which is that if you look at the new jobs that are being posted, it seems like AI skills are being increasingly sought after. So nearly one in four US tech jobs posted so far this year, what’s it called, newly listed, if you will, asking for employees to have some kind of AI skills. That’s from UMD-LinkUp AI Maps. That share’s even higher for specific categories, 40% for information related jobs, close to 30% for finance, professional services, retail education, etc. Even manufacturing was at like 20% and they’re asking for some AI skills.
(21:45):
However, that’s newly listed. If you look at all jobs, AI related listings represent just a fraction of the overall, the old and the new. I think it’s about 1% of all job listings. So it does feel like we’re being hit with this tidal wave of AI jobs or AI being listed in the job description of every job that you see now. But in the grand scheme of things, it’s not as big, as significant, as maybe we think.
(22:15):
All right, that’s where we’ll leave part one of today’s episode. Thank you so much to my guests for today. Thank you to Henry.
Henry Powderly (22:23):
Thanks Marcus.
Marcus Johnson (22:23):
Thank you to Gadjo.
Gadjo Sevilla (22:24):
Thanks again.
Marcus Johnson (22:25):
Thanks to the whole editing crew, Victoria, John, Lance, and Danny. Thank you to Stuart who runs the team, and Sophie who does our social media. And thanks to everyone for listening into the Behind the Numbers show, an eMarketer video podcast made possible by Trax. Sarah will be back Wednesday with the Reimagining Retail show and me, Gadjo, and Henry will be back Friday, part two of using AI at work, our two-part episode talking about how businesses are using it and some tips for using AI at work.