Around the World: ChatGPT and other generative AI platform adoption across regions

Audio by Bill Fisher, Carina Perkins, Paul Briggs, and Man-Chung Cheung | Feb 24, 2023

Tinuiti is the largest independent performance marketing firm across Streaming TV and the Triopoly of Google, Meta, Amazon, with over 1,200 employees and $3 billion in digital media under management. Tinuiti has industry-leading expertise in search, social, Amazon and marketplaces, addressable TV and mobile apps, Lifecycle Marketing, and more.

Episode Transcript:

Bill Fisher:

Hello, everyone. Welcome to Behind the Numbers Around the World, an eMarketer of podcast made possible by Tinuiti. It's Monday, February the 27th, and I'm your host, Bill Fisher. It's my pleasure to welcome you all to Around the World with ChatGPT and other generative AI platforms.

Welcome, folks, to a Behind the Numbers show that takes you around the world, looking at what various countries are doing in the worlds of commerce, media, and advertising. Each month, I give you a global news recap, then I speak with a few of our regional experts to get their take on the main theme for today's show, which this time centers on the buzz around generative AI and particularly Chat Gpt at the moment. We will be asking which country is seeing a boom in generative AI startups.

Carina Perkins:

The UK government is really keen on a technology-led development and really keen on innovation, and so in the future might have a bit of a lighter touch on regulation, potentially, than the rest of Europe.

Bill Fisher:

How are brands in Canada using generative AI in their campaigns?

Paul Briggs:

We're seeing brands really dig into it a little bit. I think it's very experimental at the moment, but there's a lot of interest in brands in Canada.

Bill Fisher:

And is China finally ready to start rocking the generative AI boat?

Man-Chung Cheung:

They want to get in front of these issues that they perceive will happen in the future before really start releasing this technology into the market.

Bill Fisher:

Before we do anything else though, let's give you this month's culture shock. This is where I take you to various countries around the world and regale you with a quirky culturally-specific norm. And this month, I'm going to take you to a few countries, actually, and ask, "Hi, how are you?" Because depending on where you are, the question and answer might be rather different.

In Thailand, some parts of China, and South Korea, for example, when you greet someone, it's very common to ask them, "Hi, have you eaten?" And the polite response, no matter your level of hunger is to say, "Yes, I have. Thank you very much for asking." The greeting norms in Poland, meanwhile, are a little more brutal. Ask a Pole, "Hi, how are you?" and you could get a response ranging from "I'm fine," to "It's been better," all the way to "I'm feeling pretty bad, actually." They're quite literal and very honest. Meanwhile, if you are in East Central Africa and you ask a Swahili speaker how they are, schedule in some time as a long and in-depth conversation is very much expected. You did ask after all. Anyway, I'll be saying hello, how are you, to my guests in a moment. But before I do, it's time for three in three.

I have three minutes to cover three interesting and related news stories we've seen in Around the World towers this month. The time is set. Let's go.

We're starting with Alphabet's plunge on the US stock market thanks to a European fact. At the beginning of the month, Google announced its response to Chat Gpt. In a promo for the new product called Bard, it was asked to respond to the question, "What new discoveries from the James Webb Space Telescope can I tell my nine-year-old about?" The chat bot gave a number of responses, one of which was that the JWST was used to take the very first picture of a planet outside the earth's solar system. Thing is, that was wrong. UK news brand Metro explains why.

Speaker 6:

The first pictures of exoplanets were taken by the European Southern Observatory's very large telescope in 2004, as confirmed by NASA. The error was spotted hours before Google hosted a launch event for Bard in Paris, where a Google senior executive touted Bard as the future of the company. Google's launch event...

Bill Fisher:

This error wiped more than a hundred billion dollars off the value of Alphabet as the markets thought hard about Bard.

Next, we're going to China, where Baidu has been the latest to enter the generative AI race. Perhaps surprisingly, generative AI hasn't moved as quickly in China as elsewhere around the world, indeed Chat Gpt isn't available in the country, but that doesn't mean some of the big industry players haven't been developing platforms. And Baidu is the latest to throw its hat in the ring.

The company announced earlier this month that internal testing on its Ernie chatbot would be complete in March, after which the software would be made available to the public. But China has adopted something of a wait-and-see approach to generative AI, as explained by Ronald Keung of Goldman Sachs talking to CNBC International.

Ronald Keung:

And therefore, we do think now it's still the early time for our US team. We do think generative AI is a next driver of our industry innovation. And in China's case, we are still in the process of that and likely testing phase for a few players for the next few months before any official launches.

Bill Fisher:

Baidu's announcement caused the company stock price to spike by 15%. I wonder what Ernie will be able to tell me about the James Webb Space Telescope.

And finally, we're in the UK to look at some of the legal implications that generative AI might throw up. At the end of last month, Getty Images issued a formal notification of impending litigation in the UK against text-to-image AI platform Stable Diffusion. Technology editor at France 24, Peter O'Brien, explains more.

Peter O'Brien:

There's also the legal battles between AI firms and content creators which are ramping up, the most famous of which probably Getty Images against Stability AI. They're suing Stability for scraping Getty's images from the internet to train its image generator, Stable Diffusion. And [inaudible 00:06:16]...

Bill Fisher:

This is the tip of the iceberg, of course, particularly in Europe where GDPR would seemingly render much of what Chat Gpt has done as illegal. After all, did you give the software permission to read, reuse, and profit from information about you? Chat Gpt owner, OpenAI, is not particularly open in as much as that. There is no transparency about what they do with people's private data, and there certainly isn't any explicit opt-in principle, which is required under GDPR. Expect some more legal challenges incoming. Could generative AI become degenerative AI? Anyway, that's my three in three this month.

And now, I have another three for you, three of our brightest minds on the international desk at Insider Intelligence. First up, it's our research analyst covering China and Japan. He's Man-Chung Cheung, and I should probably say hi, have you eaten? Right, Man-Chung?

Man-Chung Cheung:

Yeah, and I would say, yes, I have eaten. That means I'm doing well.

Bill Fisher:

Great. Next up is our UK-based retail specialist and senior analyst Carina Perkins. And I will say to you, Carina, good day.

Carina Perkins:

Good day, Bill. And I have also eaten. Thanks.

Bill Fisher:

Great. So hi, as it happens. And finally, we have our principal analyst for Canada. It's Paul Briggs. And for you, Paul, I'll say hello and bonjour.

Paul Briggs:

Bonjour, Bill. It's great to be here.

Bill Fisher:

Great to have you. Right, we are talking about Chat Gpt today, primarily. Because there's an almighty buzz about it, but it would be really interesting to hear how this platform and other generative AI platforms are playing out around the world. Obviously ChatGPT's owner, OpenAI, is based out of San Francisco. There's lots of development in this area in the US. But let's start with you, Paul. How do things look over the border? You're a close neighbor. Is generative AI playing out in a similar fashion there?

Paul Briggs:

Yeah, the Chat Gpt moment that's happening around the world is certainly happening in Canada as well, and several brands have dipped their toes in the water to try to execute campaigns using Chat Gpt to generate interest from some of their consumers. So for example, Taco Bell Canada used the generative AI to write a review of its Mexican pizza and made that available in its social channels. Another example is from Unilever Canada and Hellmann's Mayonnaise. They used Chat Gpt with their influencers on TikTok to generate recipe ideas for leftovers over the holidays. So we're seeing brands really dig into it a little bit. I think it's very experimental at the moment, but there's a lot of interest in brands in Canada.

Bill Fisher:

And the paid-for GPT tier is available in Canada as well, right?

Paul Briggs:

That's right. Well, it's formally has not been announced from OpenAI that it's available in Canada, but there's been several reports in social channels that users in Canada have been using the Chat Gpt Plus program. And for me, when I log into my free Chat Gpt website, there's an option for me in Canada here to upgrade. So it appears that it is available, although they have not made a formal announcement outside the US.

Bill Fisher:

Yeah, same for me. The option's there, although it gives me a dollar price. $20, right? Is that what it's saying for you in Canada?

Paul Briggs:

Yeah, US dollars. 20 USD. Yep.

Bill Fisher:

Yeah.

Paul Briggs:

Yep.

Bill Fisher:

Interesting. So Carina, Chat Gpt available in the UK. Potentially, the Chat Gpt Plus option is available. There's a bit of a hotbed of generative AI activity in Europe, though, right? In particularly one country. Tell me about that.

Carina Perkins:

Yeah, so I think the UK in particular is quite a rich breeding ground for generative AI startups. I saw some analysis by Sifted that said it had about 50 companies working in the space, and that was next to Germany and second place that only had 17. Most of them are headquartered in London. From the looks of it, there's quite a wide range of different things that they're doing.

So the Stability AI has an image generation tool. PolyAI creates human-like voice assistance for customer service teams, and Synthesia is an AI-powered video maker. So there seems to be quite a lot of innovation in the UK, and I'm not sure why that is so far ahead of Europe. I guess one thing that's probably quite interesting here is that the UK government is really keen on a technology-led development and really keen on innovation, and so in the future might have a bit of a lighter touch on regulation, potentially, than the rest of Europe.

Bill Fisher:

Is that a Brexit benefit?

Carina Perkins:

Depends whether that's a good thing or not.

Bill Fisher:

Okay, we're going to come back, actually, to Europe and GDPR and things like that. But before we do, I just want to cross the world to China because as mentioned in one of the news stories there, Man-Chung, China's taken this wait-and-see approach. Why do you think that's played out the way it has in China?

Man-Chung Cheung:

Yeah, I think that you can sort of put it that way, although there are at least 20 tech company has since announced their own sort of Chat Gpt initiative. But I think the way the Chinese government has been approaching AI development is that it's taking a cautious approach. It wants to get in front on the regulatory side to make sure that everything is safe and that national security and public interest are protected. It wants to make sure that it has this deep [inaudible 00:11:49] regulation, which I think is the first of its kind, which centers around disclosure, user consensus and transparency. So they want to... Seeing how in the past, we've seen issues with social media, and the way they look at it is that they want to get in front of these issues that they perceive will happen in the future before really start releasing this technology into the market. But yeah, I think in the coming months we'll start to see more and more of these China-based Chat Gpt being released into the market.

Bill Fisher:

Yeah, and it's a very regulated market. So I guess, would it be fair to say that some of these generative AI platforms, or the companies behind them, are making sure they've got all their ducks in a line so that they adhere to the tight regulation that there is in China?

Man-Chung Cheung:

Yeah, I mean we talk about censorship here. There are certain information that obviously the government don't want public to know about. So yeah, it's fair to say that for sure.

Bill Fisher:

So let us get back to regulation outside of China. Obviously, again, as one of those stories I mentioned at the top of the show about GDPR and how Chat Gpt and other generative AI platforms may fall foul of this kind of regulation. I mean, let's have a bit of a free-for-all here. We don't have to talk about European regulation, but how do you guys see the longer term future of these platforms when they are on the face of it, scraping the internet for information about people that they haven't given consent to be shared?

Paul Briggs:

I can speak to what's happening in Canada in terms of the regulatory sort of machinations that are starting to consider the impact of AI tools across the internet. So Bill C-27 is what it's called in Canada, it's working its way through the House of Commons. It's going to take another year or two to get past. It's really the name is Bill C-27. It's actually called Digital Charter Implementation Act. And there's a whole section in that regulation of proposed regulation that covers AI and how AI is deployed and its impact on a wide range of internet centric concerns. So that is early days and I think as these generative AI tools develop and become applied across industries, I think the regulation will become more into focus, not only in Canada but in other regions as well.

Bill Fisher:

And Carina, you had something to add?

Carina Perkins:

Yeah, I was going to say in the UK they set out their proposals for regulating the use of AI in July last year. And from what I understand, it's looking at taking a more risk-based approach than the EU which has set out its draft AI act. And that normally means that rather than kind of blanket regulating something, it will assess by risk. So it's normally a slightly lighter touch when we see that kind of approach applied. But I don't think that it's specifically talked about generative AI, which obviously has some of its own risks, but what we also have in the UK and the EU, which is already in place like you said Bill, is GDPR in the UK we have our own equivalent of that. And it sounds like generative AI Chat Gpt could potentially fall foul of that in a couple of ways.

Like you said, the problem is that it's not exactly clear how the data's been collected, but if things like blog posts, reviews, comments and articles were scraped that could be used to identify people and those people weren't informed or asked for their consent, that would be a violation of those rules. And I think another really interesting thing that people need to think about with generative AI is that when they're actually using the tool, they are potentially then giving it data on themselves or their company, which would then basically be incorporated into that tools learning and potentially be incorporated into responses that it gives in the future. So that's another thing that people need to think about, especially if companies are giving, sort of asking any questions that might relate to sensitive client data.

Bill Fisher:

Yeah, lots to think about on that front. Okay, right. Fantastic conversation, but now it's time for a new segment that we're going to try out. We might end all of our Around the World podcast with this segment. Who knows? We'll see how it goes. It's what I'm calling our recap stats quiz.

So we cover a specific topic or topics in each of these podcasts. What I'm going to do in this segment is recap them and ask our experts a few related stats questions. It should become clear. There's no prize. This is all about bragging rights, it's quickfire. We only have three questions and they're all going to be about Chat Gpt of course on this occasion. Now there aren't that many stats around about Chat Gpt given it's so new, but there are a few things that I've found out. So let's see how we go here.

We're going to start off relatively easy. So Chat Gpt was launched to the public in November 2022, but how long did it take for the platform to reach 1 million users? Now these are all multi-choice guys. So your options are three days, five days, or seven days. How long did it take for the platform to reach 1 million users? Three days, five days or seven days?

Man-Chung Cheung:

Seven days.

Bill Fisher:

Oh, Man-Chung straight in there with seven. I was going to go alphabetical, but why not? Let's go opposite, Paul.

Paul Briggs:

Yeah. I was going to say seven as well.

Bill Fisher:

Carina?

Carina Perkins:

Okay. I'm going to say five just to be interesting.

Bill Fisher:

You're going to be interesting and you are going to be correct as well.

Carina Perkins:

Yes.

Bill Fisher:

Five days. Fantastic, well done. And I think that's the fastest speed that any app or platform is reached 1 million users. Fascinating. Next up, we are currently on version three of Chat Gpt. It's actually 3.5, technically. This was trained on 45 terabytes of data and its model had 175 billion parameters. That's up from 170 million in V1. But how many parameters are expected for V4, which is expected to launch sometime later this year? Here are your options. 1 trillion, 50 trillion or 100 trillion.

Paul Briggs:

I'll go with 50.

Man-Chung Cheung:

I'll go with the highest, 100.

Carina Perkins:

I'll go with one.

Bill Fisher:

I like that there's variety in the answers here. That's great. You're not just sheep following each other. Man-Chung, you're right. A hundred trillion if you can believe it. I mean it just shows the rate of development of this thing scary right? Okay, final question. Now this isn't a stats question actually, just a bit of fun. I asked Chat Gpt to generate a made up word that would describe the future of generative AI technology. What answer do you think I got? Your options are neurocynthia. I'm not going to spell these. Zop electronics, which is a particular favorite of mine. Or IntelliFizz.

Carina Perkins:

Can I go first and choose IntelliFizz please?

Bill Fisher:

You can.

Paul Briggs:

Bill, I will go with the one that started with neuro synsice...?

Bill Fisher:

Neurocynthia. Cynthia.

Paul Briggs:

Ah, there is, I like that one.

Bill Fisher:

What you going for, Man-Chung?

Man-Chung Cheung:

There's no way I would know the answer. So I'm going to take the one that no one picked.

Bill Fisher:

IntelliFizz?

Carina Perkins:

No, that was mine.

Bill Fisher:

Oh, sorry. Which did we not say? Zop Electronics.

Paul Briggs:

Yeah.

Man-Chung Cheung:

Let's go with that.

Bill Fisher:

That's my favorite. That's a good choice, right? You're all correct, because I got it to regenerate and it came up with all three of these, which are bizarre words. I'll tell you why it chose one of them. IntelliFizz. So in its own words, I chose this word because it sounds playful and bubbly, which could represent the exciting possibilities and potential for growth in the field of generative AI models. It would say that wouldn't it, right? Let me just tot up the totals here. Now there are only three questions, so it's always likely we're going to have a tie, but don't worry, I've got a tie-break in the locker if needed. And we are going to need a tie-break because Carina and Man-Chung have tied.

So this question is just for you two. I did a Google search for the word neurocynthia. It was the one that I thought sounded more like a real word. How many search returns did I get? So just pick a number and the closest wins.

Man-Chung Cheung:

I give it 7.

Carina Perkins:

24.

Bill Fisher:

Oh, you're both in the right ballpark actually. I was expecting some of you to one of you to go a bit higher. It was actually three. So we have a winner and it is Man-Chung.

Man-Chung Cheung:

Thank you [inaudible 00:20:21].

Paul Briggs:

Well done.

Carina Perkins:

It's well done, Man-Chung.

Man-Chung Cheung:

It's my lucky day.

Bill Fisher:

You know what? I like this because unfortunately it's time to call end to today's show, but we end with a winner and that that's you Man-Chung. So well done. It's been a blast. Carina, thank you for joining us today.

Carina Perkins:

Thanks for having me, Bill.

Bill Fisher:

Paul, it's been a pleasure as always.

Paul Briggs:

Thanks Bill.

Bill Fisher:

And Man-Chung, thanks to you too and congratulations.

Man-Chung Cheung:

Yeah, fun as always. Thank you.

Bill Fisher:

And thanks to all of you for listening in today to Around the World, an eMarketer podcast made possible by Tinuiti. Tune in tomorrow for our Behind the Numbers Daily Show hosted by Marcus. And if you want to ask us any questions at all, you can of course email us at podcast@emarketer.com. I hope to say, "Hi, how are you?", to all of you next month for another edition of Behind the Numbers Around the World. Bye for now.