The Daily: The speed of the generative AI revolution, how AI can improve watching football, and Roblox's AI assistant

On today's podcast episode, we discuss the speed of AI adoption, what to make of ChatGPT's engagement lull, and how Nvidia is shaping AI's future. "In Other News," we talk about how Roblox's AI assistant could transform gaming and how Amazon is using its neural network to improve the NFL's game-watching experience. Tune in to the discussion with our analysts Jacob Bourne and Gadjo Sevilla.

Subscribe to the “Behind the Numbers” podcast on Apple Podcasts, Spotify, Pandora, Stitcher, Podbean or wherever you listen to podcasts. Follow us on Instagram

Episode Transcript:

Marcus Johnson:

This episode is made possible by Awin. Unlock unlimited marketing opportunities that reach consumers everywhere using Awin's affiliate partnerships platform. Choose which partners best match your marketing objectives, control your costs by defining how you pay, and customize your affiliate marketing program using Awin's tech to mirror your unique goals, whatever they may be. Visit awin.com/emarketer to learn more and get started today.

Jacob Bourne:

The generative AI sector is actually going to grow an average of 36% every year until 2030. And I think what's behind that is that we're going to see this technology continue to advance. And so as something bigger and better comes out, you're then going to see a fresh wave of excitement.

Marcus Johnson:

Hey gang, it's Thursday, September 21st. Jacob, Gadjo, and listeners, welcome to the Behind the Numbers Daily: an eMarketer Podcast made possible by Awin. I'm Marcus, today I'm joined by two folks. Let's meet them both on our connectivity and tech briefings. One of them is our senior analyst based on the West coast in California, it's Jacob Bourne.

Jacob Bourne:

Hey Marcus, hey Gadjo. Thanks for having me.

Marcus Johnson:

Hey Fella. And the other chap is one of our senior analysts on that very team. He is based on the East coast out of New York. It's Gadjo Sevilla.

Gadjo Sevilla:

Good afternoon, Marcus. Hi, Jacob. Happy to be here again.

Marcus Johnson:

Hello, hello. So gents, no fact of the day today. Instead, it's time for our surprise question giveaway. I brought it back last month and then immediately forgot about it. So I'm so sorry to the people who did write in with your answers. We just forgot to tell you one. Well, how does it work first? Well, every other Thursday I read a trivia question. All you have to do is either email us or send us a message on Instagram with the correct answer to the question and also include a screenshot of a review of this podcast. It can be an old one, it doesn't have to be brand new. And you can win a free Behind the Numbers branded water bottle. So correct answer plus a screenshot of your review equals free water bottle. The last question we asked, now many weeks ago, was which US president was a licensed bartender? Which US president was a licensed bartender? Do you guys know who? Any guesses?

Jacob Bourne:

Oh, that's a tough one.

Marcus Johnson:

No one's guessed right so far.

Jacob Bourne:

Adams.

Marcus Johnson:

I've asked like 10 different people, it's always different. Say again?

Jacob Bourne:

Adams.

Marcus Johnson:

No, no one said that either. Yeah, we've got another one down. Gadjo?

Gadjo Sevilla:

I don't know, Barack Obama?

Marcus Johnson:

That's who I thought it would be. I just can see him with a towel over his shoulder behind the bar just chatting it up. But it was actually Abraham Lincoln. Abraham Lincoln was a US president who was a licensed bartender before he was US president. Not at the same time, he had the country to run. And Laura from New York, you won. Congratulations to you. Stuart, who runs the team, will be sending you your free Behind the Numbers branded water bottle. So congratulations to you. Abby in Nebraska, you came in a very close second. Laura beat you by sending her answer in a few minutes before yours, so very, very close. But congratulations to Laura from New York.

Okay, the next question for another free water bottle is, here's the question, what is the oldest university in the world? What is the oldest university in the world? You can send your answers and reviews as well. Send an answer plus your review of this podcast to podcast@emarketer.com, or you can send them into our Instagram, which is InsiderIntelligence, all one word. And in two weeks we will tell you who won that water bottle. All right, time for the real topic of today, is the AI boom already over?

In today's episode first in the lead, we'll cover where we are with the generative AI revolution. And then "In Other News," we'll discuss how Roblox's AI assistant could be a game changer and Amazon using its neural network to improve the NFL's game watching experience. Although Stuart, who runs the team, as a Giants fan, not going to do much for you. You can't really improve that game watching experience, know what I mean? Jokes early.

We start of course with the lead, we're talking about the AI revolution, whether it's over already. Sara Morrison of Vox recently questioned whether the AI boom is already over, pointing out that generative AI tools are generating less interest than just a few months ago. She writes that recent reports suggests that consumers are starting to lose interest, saying the new AI powered binging search hasn't made a dent in Google's market share, ChatGPT is losing users for the first time, and bots are still prone to basic errors that make them impossible to trust. Gadjo, is the AI boom already over? Do you agree with this premise?

Gadjo Sevilla:

Not over, but it could very well be that it's the initial interest in AI is starting to slow. So a recent IDC study reported that 57% of executives believe that interest in generative AI will fade. Again, and that's through time. So I think we're starting to see maybe some saturation. No new consumer AI products have come out since the initial bar being ChatGPT releases prior to the summer. So I think that could just be somewhat a sense of fatigue from consumers.

Marcus Johnson:

So I mean, Jacob, as Gadjo mentioned, no new AI products. We've already had a number rolled out in the last, what, 10 months or so. Aside from that, why this slowdown?

Jacob Bourne:

Well, first of all, I want to say there haven't been any more blockbuster AI tools coming out. Google is actually planning one this fall. Meta is planning one early next year. So I think we're going to see it pick back up. And I think there's also a difference between excitement and then demand. I think over the long haul, like between now and 2030, we're certainly going to see a robust demand, even if people get used to it and then we see kind of an ebb and flow in traffic like what we've seen with ChatGPT. Which by the way, has been attributed to students being out of school for the summer, people going on vacation.

So I think these kind of ebbs and flows are natural, but that doesn't mean that we're going to see demand drop off the cliff. Actually to the contrary, a forecast from Forrester, a recent forecast shows that the generative AI sector is actually going to grow an average of 36% every year until 2030. And I think what's behind that is that we're going to see this technology continue to advance. And so as something bigger and better comes out, you're then going to see a fresh wave of excitement.

Marcus Johnson:

So I want to get to ChatGPT a bit more in a second, but it seems as though different people are using generative AI to varying degrees. So according to the recent Deloitte report, 42% of companies are currently experimenting with generative AI, 15% are actively incorporating it into their business strategy, 24% are reading and talking about it, and a slightly fewer 17% say it's too soon to make a decision on its use in their companies. So different folks doing different things with it at the moment. There are many companies though, gents, that are planning to ban its use according to a new report by Blackberry. 75% of companies worldwide considering or implementing a ban on ChatGPT and other generative AI apps in the workplace. Why? They say security and privacy comes out as the top reason. JP Morgan Chase actually is one such company.

Jacob Bourne:

And I think it's important to realize that ChatGPT is a very general purpose tool and companies, it's not that they're not excited about generative AI, they just want something that's customized, like something that's maybe built in-house that they can really assure that they have that security concerns covered. And going back to the Forrester forecast, actually, they expect to see a $79 billion annual spending by companies on, not general purpose tools, but specialized tools. So I think the problem is not generative AI, it's that they want something very specific for their industry that really protects their security.

Gadjo Sevilla:

To add to that, we're already seeing that happen with Salesforce. They just launched last week their Einstein assistant, which still uses generative AI, but it's curated within the Salesforce ecosystem. Which means they can guarantee the security, they know how it's being used, they can learn from the prompts and the mistakes. And so I think a lot of companies are more likely to adopt those solutions which have an extra layer of protection and are also within the tools that they are already using. So that could just mean less use of models like ChatGPT in the future.

Marcus Johnson:

So let's turn to face ChatGPT again. So a recent Axios article noting ChatGPT web traffic has fallen for the third straight month and worldwide desktop and mobile website visits to the ChatGPT site fell a further 3% in August after falling around 10% in each of the previous two months according to analytics firm SimilarWeb. And so it does seem like there's a bit of a lull in ChatGPT's engagement. As you mentioned, Jacob, school's been on a break and US ChatGPT traffic in August ticked up in concert with kids going back to school. So we saw usage drop as kids stopped going to school and now it's gone back up now they're back in school. We've mentioned other generative AI rivals as well. What else do we make of this drop in engagement and moving forward, you've mentioned, Jacob, that we're going to see a new generative AI solution from Google in the fall, Meta coming shortly after that. Is ChatGPT likely to stay out in front of everyone else in terms of usage?

Jacob Bourne:

Yeah, I think it's doubtful. I mean there's a lot of competition and even though the drop in usage is linked to students, I think we can safely say that the kind of new car smell for ChatGPT has worn off. Nothing generally stays viral forever. And I think that there's going to be new competition. Google is putting a lot of resources into building its Gemini model, which it plans to release very soon. And then of course there's Meta with its open source models that allow other AI developers all around the world to build their own apps off of it and to do so for free.

So I think there's a lot of competition that OpenAI is facing with ChatGPT. They're really under pressure to get GPT-5 out the door, the successor to GPT-4, as well as they're also working in open source model themselves. But again, in terms of looking at overall AI demand, even if one platform or tool kind of fades into the background and is not getting as much traffic, I think overall what we need to look at is the sector as a whole and I think demand, I expect to be very robust.

Marcus Johnson:

So looking at the sector as a whole, so the McKinsey Global Institute recently wrote a report about adoption of generative AI applications. It believes that if you assume steady improvement in currently known technology and take future breakthroughs off the table, mainstream adoption of generative AI will sit between eight to 27 years.

Jacob Bourne:

That's pretty broad.

Marcus Johnson:

Eight years would get you just past 2030, 27 years would get you 2050, somewhere around that as your backstop. So I mean, how fast is the generative AI revolution actually coming? Because that's quite a span. Where do you guys expect it to be on that timescale?

Gadjo Sevilla:

I think we're already seeing a lot of improvements, like currently. So for example, the digital art created by some of the generative AI tools, the quality and the realism, that seems to have increased exponentially in just a few months. There are also services that could deepfake your voice, create videos. I mean in terms of applications, we are already seeing vast improvements. So I think in general it will grow exponentially. It's just how it's reflected or how it's used, and that's really the key there. Adoption is going to be quite specific to certain aspects of it. While overall it may not look like a very big jump, we're already seeing quantum leaps in various minor industries.

Jacob Bourne:

I 100% agree with Gadjo on that. And speaking of quantum leaps, the other thing to look out for is that other technologies like the Metaverse and quantum computing, when they mature, are going to kind of have a fusion, a synergy with AI that's going to further amplify the boom, the AI revolution. And I think we're already seeing a lot of evidence that the revolution is underway. Certainly there's been workforce disruption, which is one of the, I guess, symptoms of the AI revolution in industries like marketing, tech, and healthcare. And I think also average people are getting use out of the technology as well, including being able to kind of diagnose illnesses in a way that they could never quite do on Google search. Now whether or not that's a good use case or not is debatable, but I certainly think that it's giving people access to tools and information that they didn't previously have.

Marcus Johnson:

We just talked about that on Tuesday, actually. Our Tuesday episode about Dr. ChatGPT and saying that, yeah, it doesn't need to be as good as a human physician, it just needs to be better than maybe a search on WebMD or on Google or on TikTok and all the other places that people go as well as speaking to their doctors. So we actually have some forecast numbers on this. Our wonderful forecast team has broken out some figures on generative AI users this year, 78 million people we expect to use generative AI. That is once a month. So using generative AI once a month, 78 million, close to 80 million people this year, 100 million next year. If you zoom in just on the ChatGPT folks, then we expect over 60, six zero, million Americans to use ChatGPT at least once a month and close to 80 million next year.

Just seems there's more and more people using ChatGPT. The time spent on ChatGPT is dropping a bit though, declining monthly since March from an average of nine minutes to about seven minutes in August. So it's not come way down, but it has come down somewhat. Some of that lull could be attributed to kids not being in school over the summer, as Jacob mentioned. This number's according to Reuters. Let's end gents by talking about a powerhouse in the space. A company a lot of folks may not be too familiar with. This is one company that certainly will help with generative AI adoption. It's called Nvidia.

So tech giant Nvidia says sales reached record levels after more than doubling as demand for its AI chips took flight. So the company saw revenue jump to over $13 billion last quarter, which is amazing. A lot of tech companies do make a lot of money, but as Scott Rosenberg and Ina Fried of Axios point out, doubling revenue isn't unheard of for high growth companies. Doubling it at this scale is much, much rarer. And so this strong performance was driven by Nvidia's data center business, which includes AI chips. As the BBC notes, Nvidia was originally known for making the type of computer chips that process graphics, particularly for computer games. So gents, what are your thoughts on Nvidia and how they're shaping the AI world?

Jacob Bourne:

Yeah, Nvidia made a strategic bet that paid off massively. I mean, it basically had its GPUs, graphics processing units, that it had for the gaming field and it was a leader in that sector, and it discovered that it was really great for AI workloads. So it took that framework and basically built on it and came out with two really powerful chips, the A100 and the H100, which are basically the foundation for the generative AI sector boom that we've been seeing it. In fact, if it wasn't for those chips, we might not be seeing what we're seeing right now with AI like ChatGPT. It's really made this advanced AI model training and deployment possible, and Nvidia is riding the wave of that certainly and is certainly one of the biggest driving forces behind the technology. And we might continue to see that from Nvidia into next year. They're planning a more powerful chip release. But of course there's also rival chip makers that are trying to break into the AI sector as well.

Marcus Johnson:

I mean they've come out of nowhere. In 2021, Nvidia was about 50% the size of Intel in terms of the amount it made from its chips business. Then in 2022 it was about 80% as big. And then Q2 of this year, Nvidia became 250% as large in terms of how much it made in Q2 compared to Intel. Jenna Ross of Visual Capitalist had a really good piece on this just showing the side-by-side of these companies and how quickly it has grown. The New York Times says Nvidia is said to have over 70% of the market share for AI chips. So Jacob, to your point, yeah, if Nvidia isn't doing this well, then this generative AI boom really just can't happen. Gadjo, What do you think the folks need to know about Nvidia and how they're pushing AI forward?

Gadjo Sevilla:

Yeah, so there's serendipity involved here because those chips originally were for gaming, they've been used in cryptocurrency as well.

Marcus Johnson:

Driverless cars, too, correct?

Gadjo Sevilla:

As well, yes, but these are mostly...

Marcus Johnson:

To a lesser extent.

Gadjo Sevilla:

Yeah, these were mostly used to farm Bitcoin and since that kind of went bust, the AI revolution sort of came right at the right time for them because they already had the template for making the types of powerful chips that could serve AI functions. So it really did work out for them, and I think they're going to be in the lead for a while. There's really no one that I know of that can kind of scale like they can at this point. So they do have that lead. They're constantly improving on their products and they also understand the software behind running those chips. So I think it's all good for the near future.

Marcus Johnson:

Yeah, one statistic that shows that it's all good for the near future, Nvidia made $6 billion in profit in Q2, which is good. A lot of tech companies can make a sizable amount in profit. That was up over 800% year-on-year. So it's $6 billion in pure profit. And then another figure this year, Nvidia's stock market value jumped over $1 trillion as its shares more than tripled in value. That made it the fifth publicly traded US company as it joined the trillion dollar club along with Apple, Microsoft, Amazon, and Alphabet. I believe it also puts it not just in the top five in terms of value in the US, but it's also in the top 10 worldwide.

Gents, let's skip the halftime report. We'll go straight to the second half of the show today. In other news, how Roblox's AI assistant could transform gaming and Amazon using its neural network to improve the NFL's game watching experience, just not for Stuart. Story one: Gadjo, in a recent article, you note that Roblox has introduced Roblox Assistant to help users create virtual worlds. The new assistant helps creators learn to use Roblox and can act as a tutor through back and forth conversation. But the most important sentence in your article, Gadjo, about Roblox's AI assistant and how it could transform gaming is what and why?

Gadjo Sevilla:

The most important sentence is we will continue to see companies offer tailored AI solutions to better manage and control its use on their platforms. So I think like we discussed previously, this could lead to less reliance on third party large language models and wider democratization of AI use for specialized industries. In this case, it's for game development. So customizing AI solutions could eclipse standalone AI large language models like ChatGPT by giving businesses the opportunity to cash in on AI.

Marcus Johnson:

And Roblox just doesn't seem like as big a platform as it is. When you actually look at how many people use it, it's not a small platform. Roblox has around 60 million daily active users, 200 million monthly active users according to the company, and not just for really young kids. Over half the users are over the age of 13 years old. A good amount of them are young teens in that 13 to 16 range.

Story two: Jacob, you recently wrote that Amazon Prime Video will roll out new AI powered broadcast features for Thursday night football, which it now has the rights to, to enhance the sports fan's viewing experience. You explained that a feature called Fourth Down Territory will help the audience understand how players make fourth down decisions, and there's a field goal target zone feature, which is a screen indicator giving you the probability of a kicker making a field goal as well as other features. But Jacob, the most important sentence in your article on Amazon using neural networks to improve how people watch NFL games is what and why?

Jacob Bourne:

So part of the allure of live entertainment is that viewers enjoy the thrill of not knowing what'll happen next. And I think the takeaway here is that there's a myriad, probably countless use cases for AI and some of them are better than others. And so I think companies like Amazon really should be careful about and really think critically about how they're deploying it. I mean, certainly integrating AI into Thursday night football is going to have the factor of novelty there where people will want to see what it's all about and maybe some of the ways they integrate it will be useful, but I think there's also a risk of kind of eroding the viewing experience if it gets in the way that predictive element of AI kind of gets into the way of the anticipation. That's kind of the point of a live viewing experience. You don't know what's going to happen next. You want that anticipation. If AI is kind of always there telling you what might happen next, it could kind of take away from that.

Marcus Johnson:

That's a great point. I was joking about Stuart, who runs the team, being a Giants fan and how painful that viewing experience is, but it wasn't last Sunday when they had a historic comeback against the Arizona Cardinals, and had you been watching that game, AI could have said, oh, the Giants have a 2% win probability and maybe some people would've shut the TV off and they shouldn't have because they stormed back and ended up winning 31 to 28. Now, will it be the only win of the season? Probably. But they still won that game, so congratulations to you.

That's all we've got time for for today's episode. Thank you so much to my guests. Thank you to Gadjo.

Gadjo Sevilla:

Thank you very much.

Marcus Johnson:

Yes indeed. Thank you to Jacob.

Jacob Bourne:

Pleasure being here, Marcus. Thanks.

Marcus Johnson:

Thank you to Todd who's editing today's show, filling in for Victoria while she is away for a few days. Thank you to James who copy edits it, and Stuart who runs the team. And thanks to everyone for listening. Please send us in your answers to what is the oldest university in the world. If you want to win that free water bottle, you can email in your answers and reviews to podcast@emarketer.com or head to our Instagram at InsiderIntelligence, all one word. What is the oldest university in the world? Answer that, plus send us a screenshot of your review to win the free water bottle. We'll see you tomorrow hopefully for the Behind the Numbers Weekly Listen: an eMarketer Podcast made possible by Awin. Congratulations to Laura from New York.