Episode Transcript:
Marcus Johnson (00:00):
This episode is made possible by Round L. Partner with Round L and reach over 165 million guests who look to target for joy and inspiration. Together they'll design curated media solutions that are a seamless extension of the target experience, and it's all backed by unparalleled first party data and of course measurements. If you want to know more, you can head to round L-R-O-U-N-D-E l.com.
Dan Van Dyke (00:30):
It's important as we all raise to realize those efficiency gains, to also see some of the risks of sharing data that shouldn't be shared and to keep our eyes open about working with AI players whose valuations seem to imply that investors believe that they'll disrupt a lot of industries.
Marcus Johnson (00:52):
Hey gang, it's Tuesday, June 4th. Dan Henry, and listeners, welcome to the Behind Numbers Daily, an eMarket podcast made possible by Round Dale. I'm Marcus. Today, I'm joined by two gents. Let's meet them. We start with our VP of Gen AI based in New York City. It's Dan Van Dyke.
Dan Van Dyke (01:11):
Hey Marcus. Good to be here.
Marcus Johnson (01:12):
Hey, fella. We're also joined by our SVP of Media Content and Strategy based in Maine. It's Henry Powderly.
Henry Powderly (01:20):
Hey Marcus. Thanks for the invite.
Marcus Johnson (01:21):
Hey, fella. Yes, sir. Welcome to the show. First time we start with the fact of the day our brains produce as many as 50,000 thoughts per day according to the National Science Foundation. That would be about 40 a minute. I don't know if mine does
Henry Powderly (01:42):
Half of them come right before you're trying to fall asleep, right? Yeah,
Dan Van Dyke (01:46):
Pretty
Marcus Johnson (01:47):
Much all of them.
Dan Van Dyke (01:47):
Yes. I feel like if I hit 40 a day, that's the good day.
Marcus Johnson (01:52):
It's solid pace. Almost all of them. 95% are repeated thoughts and 80% of them are negative unless you're English and then they all are.
Dan Van Dyke (02:04):
How would they even do that study? I
Marcus Johnson (02:06):
Know. How would they determine
Dan Van Dyke (02:07):
80% or negative?
Marcus Johnson (02:09):
This may or may not be made up anyway. Today is not today's real topic, how gen AI is changing how we work.
Marcus Johnson (02:22):
So just the lead on Gen AI know in other news today. Let's get to it. So last December 14th, myself, Dan, and Jacob was on the show. We talked about the do's and don'ts of using genai in the workplace and how employees feel about it. We covered what employers expect their staff to know about genai, whether every company will end up with their own custom version of chat, GPT, and the do's and don'ts of using genai at work. Today we're covering some other topics related to generative AI and working with it at work. We'll start with open AI's, new GB t4 Oh, and what that can do for companies and employees. So tiny history here for folks, OpenAI just introduced its latest model, GPT-4 oh. It's twice as fast can better digest images and video in addition to text. It can interact with people by voice in real time, offering a more conversational rhythm. It can detect a person's emotions in their tone, voice, or facial expressions, and it also can remember things. So Dan, I'll start with you. What interests you most about Open AI's GP t4? Oh, in terms of what it can do for companies and employees.
Dan Van Dyke (03:28):
So what it really does better at launch is that it's a faster B, cheaper to better writer. It handles audio, it handles images better, whereas previous versions of chat GPT, like 4.5 or just the initial launch of chat GT at all g PT four were Quantum Leap. What we're starting to see now is it's more like the more recent iPhone launches where everyone is just a bit better. And the exception or where that analogy dies is that it's also a lot cheaper in this latest launch. So what chat GPT is able to do is give all its free users access to this new model as well, which is really exciting. And then also a little bit scary when you consider the hidden cost of using a free service is that you are the service. So it's a great way for them to capture a lot of data, enable a lot more use cases that couldn't have been possible in the past. But Henry, I'm curious for your experience having tried it out for a while, what do you see as the key use cases or advantages?
Henry Powderly (04:27):
Yeah, and I think you kind of hit it on the head there. It's incrementally better version of GPT than GPT-4, so it does write better. It seems to have just a little bit more nuance in the writing already. GT four I thought was pretty good at writing and kind of a leap over the 3.5 model. So I'm seeing a lot better writing. I'm excited to test out the audio input changes that I don't think are available to everybody yet that's really supposed to make the conversational interface a lot stronger. One of the things I've been reading a lot about is having the difference in working with GPTs prompting versus conversing with the ai. I think for the last few releases, it's been very prompt focused where you craft a well thought out request or a command for the GPT and you work with the outputs that way. But in terms of things like brainstorming and ideation and decisioning, you could envision a conversational interface working that much better for that. And that's supposedly what 4.0 does. Four oh does a lot better. And if you saw some of the demos that OpenAI shared with the release, it did seem to be a lot stronger. And the fact that it's reading your facial expressions too in that conversation is really interesting.
Marcus Johnson (05:37):
So Dan, whilst we're thinking about putting this episode together, I think I said to you that it seems like the idea is that Genai can give employees superpowers and there's some data from the law AI and automation as a survey. They did and they looked at the average amount of time that we spend on work tasks, and I won't go through all of the shares, but they're pretty evenly split between seven to 11%. The biggest is 11, it's reading and responding to emails, talking on the phone also has 11 entering data as well. Then it goes analyzing data in-person meetings, editing documents, managing scheduling research and reading virtual meetings, creative thinking and coding software, the smallest percent at four. What do you think this four oh version helps with the most?
Dan Van Dyke (06:23):
It's interesting that you asked that. I recently read Benedict Evans piece where he was talking about how chat g BT just doesn't have the killer use case that a lot of other technologies have. And I think the disappointing answer to your question is that it's all of those things. It helps you get better at all of those things. And to put a little bit of meat on the bone, what I would say is take whatever task you get stuck with that really disrupts your state of flow. So if it's trying to change an audio setting and you're stuck, you can turn to chat GBT and say, help me out with this annoying thing. And then you're right back in the flow of work. That's what I find most magical about the process is no longer having those tiny disruptions that add up into a lot of switching costs, a lot of frustration, and ultimately a lot of getting taken away in terms of your attention from the core task.
Marcus Johnson (07:15):
Henry, do you agree with Benedict Evans and Dan here that there isn't a killer app and that it is everything? Or has one thing jumped out to you?
Henry Powderly (07:24):
The writing has jumped out to me a little bit just because when chat GT first came out and the promise of being able to use it for copywriting or composing emails, when you first started working with the tech, it felt like it was the sixth grade writing level, and even though the drafts were workable, it really wasn't quite there. And now with four, oh, I really think we're close. There are some prompts I've given it for marketing emails based on an asset abstract that are pretty darn close to what I'm seeing in my inbox every day. And with a smaller level of editing and refinement, you really getting it from inception to delivery ready in a short amount of times. Yeah,
Marcus Johnson (08:10):
Yeah. It seems like there are things people will favorite it for versus others, but Dan, to your point, nothing is clear cut head and shoulders. This is what you're supposed to use it for. It works well for a lot of stuff.
Dan Van Dyke (08:20):
I mean, to give you a little bit better of a response, I hate when people give answers like the one I just gave. I would say that a lot of people use it to take a small prompt and accordion it out into a large piece of text. And what I would say is the most powerful use case is to take a large body of text, say it's the transcript from this podcast or it's a collection of reports and find what's most relevant. So to accordion it in that summary and analysis is the superpower as opposed to extrapolation and being long-winded, although it can do that if you want to.
Marcus Johnson (08:59):
Yeah, like we was talking about, folks are using these things at work, but employees are bringing their own AI to work in a lot of instances. And Ryan Heath of Axios was pointing this out and he was explaining that three out of four knowledge workers around the world are using genai, but many of them are hiding it from their employers. According to a new joint report from LinkedIn, it's parent, Microsoft Henry. What do you make of employees bringing their own gen AI into the workplace, BYO Genai, if you will?
Henry Powderly (09:28):
I think there's nuance here to think about because the reality is every tool that exists right now is getting AI infused into it. The platform that I use for email, for example, in the past few months now has an AI feature for proofreading and helping shorten emails and adjust tone that it didn't have a year ago. And so it's not like these are sneakily being brought into companies, but sometimes some of the very tools that people we're working with are already getting enhanced by ai. I mean, the other side of this though is certainly there's a lot of experimentation, a lot of excitement around ai, and I think you want to have people experimenting and trying new tools. The challenge is going to be for companies to keep tabs on that and to really talk through where the pitfalls are and what needs to be considered, especially when you're talking about sensitive data.
Marcus Johnson (10:20):
Yeah, it seems like Dan, one thing it does though is it makes it harder to measure the benefits of a gen AI might have for a company. Mr. Heath was pointing out because employees are using AI outside of the official company strategy and rules, bosses are struggling to measure AI benefits and adjust their AI investments accordingly.
Dan Van Dyke (10:36):
Yeah, I mean, that's an interesting point. I think there is a lot of efficiency gains that are just purely dark to us. And we could ask through survey questions to say, what other tools are you using? And maybe we get a complete answer if everybody feels comfortable sharing that information, we could ask them to say, how much time are you saving? But then it's self-reported and human memory is flawed, some including myself worse than others. And so I think the reality is that almost everybody would agree that we're saving time as a result of using gen AI tools. And I think it's important as we all race to realize those efficiency gains, to also see some of the risks of sharing data that shouldn't be shared and to keep our eyes open about working with AI players whose valuations seem to imply that investors believe that they'll disrupt a lot of industries.
Marcus Johnson (11:30):
You mentioned self-reporting in terms of efficiency gains of folks using their own gen AI for different tasks. Also note here, so folks just incredibly nervous about sharing some of this stuff. The piece was also saying of AI users surveyed half said they worry that if their employers know they're using AI to be more productive or creative, it will signal that they're replaceable. And so you want to use this stuff to help you out with your day, but you also don't want to use it too much because if you can, does it suggest that maybe it could do the entire job for you? Folks also might be less inclined to bring their own gen AI if there was maybe more training, maybe if things were more standardized at a company less than four and 10 employees. So their company provides AI training according to that link to the Microsoft report. And I imagine gents we're going to see more, at the very least, AI literacy classes at companies. Could you see that happening soon? Yeah,
Dan Van Dyke (12:22):
Absolutely. I mean, I'm currently taking a section course that we get access through Axle Springer to upskill myself on Gen ai. I'm working on trainings that will roll out across the organization with other people hoping to tap you for this. Again, Henry, because he ran such great trainings in past years. But I think from my perspective and the perspective of everybody I've spoken to who's been on the side of, let's lean into this change, you'll probably lose your job first to somebody who's using AI as opposed to losing it to AI purely. And really the thing to do to become insulated to this really disruptive technology shift that probably will result in some jobs being moved around is to become the person who's an expert at using AI to do multiple of what you could do in the past without stretching out your workday.
Marcus Johnson (13:15):
Yeah, we have run some of those classes at the company in terms of helping folks understand this technology. Henry, what have you made of some of those classes and what folks are responding well to in terms of what they want to learn and what do you think will be the next wave of teachings for folks in terms of gen ai?
Henry Powderly (13:33):
Yeah, and like Dan said, I happen to lead a couple of them here. And for the most part, I encounter a lot of enthusiasm about the tech. I mean, I do think there's a worry that everybody has that this is going to replace them, especially in the knowledge work field where we sit. But by and large, I think people are really excited about it because there are those kind of low value tasks that folks in our line of work are saddled with every day. So the premise of offloading some of that to a software that can help you move faster and focus more on the high value things that you do, I think is really exciting to people though it's a change in how we work. And so with any kind of change, there's lots of questions and I think we just need to be patient and as a team, work through those and give everybody as much hand ons experiences we can, working through all those challenges.
Marcus Johnson (14:21):
And that's looking at the individual employee and their daily tasks and how can this technology help make some of those more monotonous tasks go away and make them more efficient. But if we zoom out our very own, max Williams was interviewing Kristen Kelly, global Head of Media and MD at Accenture Song who told him 50% of our scope in conversations is talent and org change design as it relates to integrating Gen AI into a company. Henry, I'll go to you first. What on how an organization might be restructuring in terms of where it's putting its people and what teams look like as it relates to gen ai.
Henry Powderly (14:59):
And I think we still are in early days, and so the first thing I see is folks like Dan at our organization who are appointed into roles that are focused on AI and governance and training and use cases. And I guess the real question is what happens after it gets mature and AI becomes embedded in the organization and that literacy level is included? Because at that point then, do you really need a specific AI oversight or is it really just business as usual?
Marcus Johnson (15:26):
Yeah, down.
Dan Van Dyke (15:27):
Well, hopefully we still need me, but we'll see. I agree with what Henry's saying. I think ultimately AI is going to be so embedded in the processes that we use and our workflows that probably won't be distinct AI units except for within technology and engineering and data teams in that ultimately it'll be everybody's using this tech and it's as ubiquitous as the laptops that we're using right now. So it'll become boring and it'll just fade into the background.
Marcus Johnson (15:59):
Yeah. Speaking of organizational change, let's end here. So Ryan Heath of Access again had another piece noting AI shaking up corporate boards with the most aggressive companies appointing AI bots as observers to their boards to help make more informed decisions. Mr. Heath points to one of the United Arab Emirates most valuable public companies called International Holding Company has appointed Aiden Insight to its board, which is an AI observer. He explains. Aiden Insight is the persona of a tool called Board Navigator created by G 42, the Gulf Region AI company that just secured 1.5 billion in investment from Microsoft. Henry. Why do you think AI should or shouldn't have a seat on the board?
Henry Powderly (16:42):
Well, I mean, just taking a step back, I think the story is interesting and it's a little bit hyperbolic and not necessarily widespread at this point, but what it's really getting at is the value of AI and decision making that can go up at the highest level in the company with the board seat. I mean, what they're really talking about in that article, it's something in the room that has the full context of everything that's gone on about your companies that's embedded with your company values and ethics. So it can check against decisions. It has all sorts of data that can be analyzed on the fly so that you make smarter decisions. And I think that's essential for somebody as high up as a board or in the C-suite. But what's really interesting is how democratizing it's because those tools are just as available to any leader in an organization. And I think you're going to see much more of these kind of embedded agents at management levels throughout the entire company, not just the board. Yeah.
Marcus Johnson (17:35):
I mean, talking about throughout the entire company, David Feld of the New York Times had a piece titled, if AI Can Do your job, maybe it can also replace your CEO. Dan, what did you make of AI at this kind of very upper echelon level of corporate America?
Dan Van Dyke (17:49):
I think that typically the CEO role is a distillation of problems that are too difficult for everybody else to solve, and so they go up the chain and that if you were to outsource that to ai, you deserve what's coming to you, which is probably not good for your business. But then again, AI hallucinates some really good CEOs like Elon Musk hallucinate. So who knows if you actually took a hard look at the human baseline and what AI could do, how far we are from a world in which the premise of that question is not just an interesting sort of clickbait title, but something that realistically we'll be looking at as a possibility in the future. And the more interesting thing for me to think about right now is how to enable the conditions by which that could happen that Henry was describing, which is you have a knowledge graph that connects all the internal or external data that you handle on a day-to-Day basis, because otherwise that would not be possible.
Marcus Johnson (18:44):
Yeah, the knowledge part of this is really interesting because it's talking about realtime insights, right? And it's talking about taking that and pairing it with external market data as well to offer advice around, and that's a pretty powerful proposition, MD of a Mad Venture group. Steve Singh contextualize it by saying quote, it's crazy for a board to still get updates every 90 days quarterly with PDFs when predictive AI can spot a revenue issue two weeks into a quarter. I thought it was a really interesting point. I wonder, gents, if this changes the expectations in terms of the frequency of reporting, whether for Wall Street or internally shareholders might be asking for a brief biweekly or at least monthly reports as opposed to quarterly updates. The SEC required public companies should report quarterly earnings in a standardized manner for the last 50 years in 1970. So I wonder if we do an update there in terms of how regularly and more detailed people expect reporting to come in.
Henry Powderly (19:40):
Certainly. I mean, and if the feed of data is constantly updated and attached to some kind of an AI model, I mean, it's really a shift to an on-demand reporting model, or at any point, any person could request the analysis that they're looking for.
Dan Van Dyke (19:55):
Yeah, I think a lot of that can be solved by just using technology like Notion that takes the hub of information and then pushes it out into different spokes. So if you look at a company like Levels, which is Smart Health Company, their investor updates are all done through Notion, and they're a lot more regular than typical. And that's what VCs depend on, and it's really rich and it actually serves as more of the operating system of the business where you can add a glance, see all the KPIs that feed into saying how well the business is doing, and that's what they use. AI could play a small role in that, but I think there are other solves that AI could have nothing to do with that could lead to that outcome that you're describing.
Marcus Johnson (20:36):
Yeah. Seems like folks saw this coming. AI on the board, though nearly half of respondents to a 2015 World Economic Forum Tech survey predicted first AI board directors would appear by 2025, which of course is a year and a half away. I'm just waiting. I know I'm going to get the email as soon as Mattias, our CEO hears this episode saying, replace me with ai. Will you? I'm not suggesting that. Okay, this might be the last episode I do. I probably lost my job by doing this, so I got time for, but this episode, thank you so much to my guests. Thank you to Henry. Thanks, Marcus. Thank you. Sit down. Thank you. And thank you to Victoria who edits the show, Sue and Sophie, who are also part of the podcast crew. And thanks to everyone for listening into the Behind the Numbers Daily and eMarketer podcast made possible by Roundel. You can hang out with Sarah Libo tomorrow on our Reimagining Retail show, where she speaks with Becky Shilling and Zach Stambaugh and takes a pulse check on how retail has done over the last couple of months.
%20scale(0.100000,-0.100000)%22%0A%20%20fill=%22%23000000%22%20stroke=%22none%22%3E%0A%20%20%3C/g%3E%0A%3C/svg%3E%0A)
