What Marketers Need to Know About Generative AI
Digital Marketing | February 21, 2023
- Marketing execs say the hype around generative AI is real, and they expect today's digital media landscape to be turned upside-down with the rise of a new generation of advanced chatbots that summarize information on the internet without traditional search engines.
- The shift from search engines to generative AI will accelerate over the next 12 months, as Microsoft and Alphabet face a classic "innovator's dilemma" whereby they must disrupt their own search advertising businesses at the risk of being disrupted by other competitors. Meanwhile, Apple and Amazon will likely use their voice-enabled AI assistants (Siri and Alexa) to position themselves as bigger digital media players as well.
- Publishers will gradually lose web traffic from search engines, hurting their own advertising revenues, causing further strife and disruption in the media industry. Alphabet's future business model may look more like a cable company, licensing content rights from publishers (like ESPN and CNN) which is then bundled and distributed to consumers via voice- and text-enabled chatbots for a subscription fee.
In December 2022, a for-profit AI research lab named OpenAI released ChatGPT - a chatbot application that uses generative AI and machine learning to automate search and summarize information users are looking for on the internet (recipes, product reviews, math solutions, news, etc.). ChatGPT immediately captured the imagination of consumers, and within 30 days of launching, the ChatGPT website had amassed more visits than major publishers such as ESPN.com and NYTimes.com. The comparison to media companies is prescient, because generative AI applications like ChatGPT could become the new "pipes" for distributing digital content to consumers, according to marketing executives in Hibben's network.
The Current State of Generative AI Applications
In February 2023, Microsoft integrated OpenAI's advanced chatbot functionality with a new version of its search engine Bing, and Alphabet piloted an integration between Google Search and its new advanced chatbot application called Bard. Users have reported some errors with both applications, but overall consumers seem to have maintained a great deal of interest in the technology. Around the same time, Chinese tech giant Baidu released its own advanced chatbot application called Ernie.
Currently, ChatGPT and Bard are only available as text-to-text applications, but in the future we will likely see this technology integrated with voice-to-voice applications like Siri, Alexa, and Cortana. Apple is the dark horse in the generative AI space, as the company acquired more AI companies between 2010 to 2021 than Alphabet, Microsoft, and Meta. Moreover, Apple recently expanded its digital ad inventory to include podcasts, books, and maps -- suggesting Apple is looking to become a bigger player in the digital advertising market and it could use its AI technology to do just that.
“We’re probably going back more to the display advertising model." says Oded Netzer, Vice Dean of Research at Columbia Business School.
“This is both amazing and terrifying at the same [time]," thinks George Stakhov, Chief Strategy Officer at DDB Worldwide.
The Innovator's Dilemma
The macro shift in the advertising industry for the past twenty years has been the shift from print to digital -- a transition that has largely played out already. Digital currently accounts for 72% of US ad spending, up from 54% in 2019, according to eMarketer. TV accounts for an additional 17% of ad spending in the US, but no other channel captures more than 3% of spending. Over one-hundred years of data show us that total US ad spending has always equaled around 1% of US GDP, so advertising is very much a zero-sum market, and very soon digital ad players like Alphabet, Meta, and Amazon will be fighting over existing, rather than net-new, digital ad revenue.
Today's digital ad industry is consolidated between two general types of formats: search and display (which includes social and video). If you breakout social and video from display advertising, then search is by far the largest digital ad format. Alphabet is the dominant player in the search ad market with 70% share, according to eMarketer. So if generative AI applications do become the new pipes for distributing digital content on the internet, and if the providers of those applications decide to monetize those platforms with advertising, then Google will have the most to lose if the search ad business model begins to fall apart.
Microsoft CEO, Satya Nadella, told the Financial Times that "from now on, the [gross margin] of search is going to drop forever." He's also said that Microsoft will continue monetizing Bing using advertising, suggesting Microsoft is simply more willing than Alphabet to lead a race to the bottom in search ad pricing. Neither company may have a choice in the matter as it's a classic "innovator's dilemma" whereby Alphabet and Microsoft must decide whether they are willing to disrupt their own search ad business models or risk being the ones disrupted.
However, a more likely explanation for why search ad prices will decline is that generative AI will commoditize internet search queries resulting in far more inventory for advertisers. Google currently process 1.2 trillion search queries a year but voice-to-voice AI applications like Siri and Alexa have led to an additional 50 billion search queries annually. And in December 2022 alone, the OpenAI website that hosts the ChatGPT application received 667 million visits, which suggests billions of additional search queries through this new generation of advanced text-to-text AI applications. So generative AI will commoditize internet search, creating far more ad inventory, driving down ad prices.
Publishers Can't Catch a Break
Traditional media companies have had to weather a series of disruptive industry shifts the past thirty years. First, it was the shift from print to digital and cable TV to video streaming. Then it was the shift from desktop to mobile and the rise of social media. Now, they must learn to adapt to a digital media landscape where the "Big Tech" traffic cops (i.e. Alphabet, Meta, Apple, etc.) deliver publishers' content directly to consumers using generative AI applications. This will lead to declining web traffic to publishers, which in-turn will result in less ad revenue for publishers, which will cause further strife in the media industry. One of Hibben's members who is a retired technology executive, says he expects publishers to "try to block generative AI applications from using their content," and that it will lead to "further debates among the Federal Trade Commission (FTC) on whether tech companies have the right to aggregate other publishers' content, even if they ultimately generate new content from the source."
Reducing search traffic to publishers would be a huge blow to the media industry. News organizations like Bloomberg and CNBC receive 30-50% of their web traffic from search engines currently, according to SimilarWeb. The best-case scenario for publishers is that consumers continue to use traditional search engines to find content that interests them but that they also use generative AI applications to question and go deeper into those topics. For example, consumers may still end up going to ESPN.com for some general information about the upcoming Super Bowl but they may use a generative AI application to pull statistics from the last time those two teams played each other -- in this use case, publishers and generative AI can each provide complementary value to consumers.
Another way publishers may benefit from generative AI in the future is they may get paid to allow tech companies to use their content for generative AI applications. Advanced chatbots like ChatGPT are only as good as their training models which rely on massive amounts of information across the internet, including articles, wikis, forums, and product reviews. If publishers do try to prevent generative AI applications from sourcing their content, or if the FTC takes the copyright issue to court then tech companies may just end up licensing content from publishers -- especially if generative AI applications are sold as subscriptions to consumers. In this model, the tech companies almost act as cable companies, licensing content and distributing it through pipes, which in this case would be the generative AI applications. This would also encourage publishers to focus more on creating premium content and spending less time and resources on selling their own advertising.
Get our free weekly briefing.