Google's AI shift is causing a collective freak-out - Taipei Times

When Google announced radical changes to its search tool that would overshadow the page of blue links we have been used to seeing for more than a decade, online advertisers had something of a collective freak-out. The Alphabet Inc-owned company called it the biggest shift in more than 25 years, explaining that the search bar would be "completely reimagined" with artificial intelligence (AI).

Forget sponsored links, now "conversational discovery ads" would be built to sit inside AI answers themselves and potentially include a chat agent so the user never clicks through to a Web site. On the day of the announcement, an online-advertising lobbyist I know texted their contact at Google to find out what it meant for their industry. The answer was noncommittal. Even some people at Google do not seem to know where this will lead.

The update also received immediate pushback from a British regulator, which ordered Google to let Web site owners block their content from being used in its AI search features. Some publishers would see that as a necessary defensive measure, but others might fear missing out on a new route to eyeballs. That is because the technology is largely untested and its impact is unclear.

For years, companies relied on a few basic techniques to boost their chances of topping Google search results, like spamming the Web with keywords or getting other businesses to publish so-called backlinks to their Web site — a way of establishing its credibility with the search engine. Such signals were crafted to appeal to Google's PageRank system to get on the coveted first page of results.

However, the tech giant's sweeping new changes mean that over the next year or so, internet users are less likely to see those blue links. Instead, they would be interacting with Google's AI model Gemini or software agents that would make product recommendations and, eventually, even purchases.

AI overviews — the singular answers that have sat on top of search results for the past two years — have been a taste of the bigger AI takeover of search to come, and have sparked much fumbling in the dark by companies and advertisers eager to appeal to Gemini and its ilk instead of just PageRank.

The new dark art of search engine optimization is known as GEO (generative is now the first word), and it paradoxically has a more human flavor than the era of backlinks. To get noticed by AI, companies must sprinkle the Web with more conversations about themselves. A business that makes climbing shoes, for instance, stands to benefit from positive chatter about its product in published articles, TrustPilot reviews and discussions on popular forums, especially Reddit, a favored source of scraping for AI models.

When a Web user then asks ChatGPT or Google about the best value-for-money climbing shoes, the underlying AI model would base its answer on an analysis of conversations about the product on those sites, and whether overall sentiment is more positive than negative. The idea is that while this approach is tapping actual consumers, its aggregate view would be useful and less easily gamed than the star-based review system popularized by Amazon.com Inc.

"Humans tend to buy using emotion, but AI agents will be more rational," advertising agency WPP PLC chief AI officer Daniel Hulme said.

Hulme said that as chatbots become just another platform for advertising, such as television, YouTube or social media sites, so too will new channels such as AI companion apps — not made by the big labs — which can also make product recommendations. Anthropic PBC has denied that it would introduce ads into its popular chatbot Claude, while OpenAI has said ads on ChatGPT would not be woven into its answers.

However, that would not stop smaller, third-party AI tools from going down the product placement route. Hulme is building an app he describes as an AI concierge to identify potential human matches (akin to Tinder) as well as product matches.

What that means more broadly for online advertising is hard to say. At a recent ad-tech conference in London, several attendees from the industry told me that while they were accustomed to adapting to algorithmic changes wrought by Google or Facebook over the past decade, the latest AI shifts felt like the ground was moving under them. Some predicted that the global market for internet ads would shrink over the next few years as a result.

The overriding sentiment, including for Alphabet insiders, was uncertainty about how people would change the way they use Google or respond to ads in its AI answers.

Google has always run on experimentation, testing features across different cohorts of users to see what they respond to most enthusiastically. It tends to follow where the online public goes. Ten years ago, you would typically type a couple of keywords into its search and hope for the best. Now people are typing in longer, more specific sentences, another trend that changes the calculus for advertisers.

This time around Google is being led somewhere uncomfortable — for the company, as well as advertisers. Jumping on the generative AI trend has led it to partly dismantle the click-a-link business that made it fabulously rich, and analysts are split on whether that is a good idea. Some warn against the obvious cannibalization of its highly profitable model, while others say Google itself would be just fine thanks to its expanding cloud-computing business.

Goldman Sachs analyst Eric Sheridan says that Google's AI scale and infrastructure would keep it growing, even if the old search-ad business gets squeezed as AI agents take over more of the searching. In other words, advertising — about three-quarters of Alphabet's revenue, but a shrinking share as the cloud business grows — might just become a smaller piece of the pie. The ad business still pulled in about US$77 billion in the first quarter of this year, up almost 16 percent from the same period the year before.

At the same time, profits from the old advertising model are helping Google fund its big AI switch, alongside a fundraising push that takes advantage of the market's AI mania.

Alphabet is raising US$80 billion in equity, it said this week, including US$10 billion from Warren Buffett's Berkshire Hathaway, to pay for the computing power behind its new ambitions.

So while Google's evolution takes it into uncharted waters, ultimately that might be less of a problem for Alphabet than for advertisers.

Parmy Olson is a Bloomberg Opinion columnist covering technology. A former reporter for the Wall Street Journal and Forbes, she is author of Supremacy: AI, ChatGPT and the Race That Will Change the World. This column reflects the personal views of the author and does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.