Shares of Google's parent company Alphabet fell nearly 8% on Wednesday in the hours that followed Google's presentation, shaving roughly $100 billion from the company's market cap. On Thursday, shares fell another 5%.
?USD
(GOOGL)
Wednesday's event was held in Google's Paris office just a day after Microsoft (MSFT) revealed its new version of Bing, which it's integrating with OpenAI's much-discussed ChatGPT. Google's dominance in search was once seemingly untouchable, and the numbers certainly reflect as much – Google has more than 80% market share, while Bing comes in at roughly 9%.
So, why did Google's presentation raise red flags with investors on Wednesday? SEO expert Lily Ray says it's a combination of things, starting with the fact that the company, in the end, didn't really announce much that was new.
"Honestly, it [felt] like things Google's already announced before," she said. The new features that Google executives focused on in the company's Wednesday presentation included things like "search your screen" for Google Lens, which was in beta in 2021.
The presentation also included the announcement that Google is rolling out its "multisearch" option globally. The feature allows users to take a picture of a green chair, for example, then type 'blue' to find the same chair available in blue for sale online. Ray's been excited about and using multisearch for almost a year.
"Multisearch is one of the most exciting things they've come up with in recent years, and I think it's going to change the way we search," she said.
Still, everyone was anticipating news about Bard that would be truly illuminating – and it never quite materialized.
"We were all waiting for Bard news, but the announcement was pretty consistent with the blog post from two days ago," said Ray. "We saw a pretty limited first view of how Bard will look in the search results, but not a lot of clarity on when or how frequently it will be triggered."
Ultimately, the presentation itself was, well, a bit un-Google-like.
"There were a couple of fumbles, and it didn't seem to be Google's most well-planned announcement," she said. It also didn't help that the presentation somewhat ended abruptly, with the video going private on YouTube for a few hours, said Ray. The video is since back up.
Google declined Yahoo Finance's request for comment.
'We were expecting to be amazed'
Simultaneously, concerns about Bard's accuracy have already come to the fore, as an example that Google used in promoting its AI chatbot wasn't exactly accurate. Though reporting from the Financial Times has suggested that the example Google used about the James Webb Space Telescope isn't "on a technical level, wrong," it's nevertheless a brutal gaffe for Google. This sort of quandary is also characteristic of the types of language models that both Bard and ChatGPT fundamentally are.
“One of the major challenges with these large language models is that they can’t say ‘I don’t know’ and instead produce a response it deems accurate," said Alex Ratner, CEO of Stanford AI Lab spinoff Snorkel AI. "And often, the response falls short of the truth. This is because, in the end, the model is as good as the data it is trained on, and a lot of the data we produce in the world is unstructured—meaning, it’s unlabeled and unclassified."
Meanwhile, what Microsoft announced yesterday is genuinely new and, from an SEO perspective, innovative, said Ray.
"For me, this announcement was the most exciting and interesting step they've taken with Bing in a long time," she said. "But like Google, they still have a lot of answer for as to how this might affect online publishers."
At its highest levels, this is a battle Microsoft going all-in on. In an interview earlier this week, Microsoft CEO Satya Nadella took a swing right at Google, saying that as searches grow more complicated, they'll get more expensive — and the high margins that have built Google's business will fade away.
Nadella's right, and Google might have to recalibrate the economics of its search business if it plans to be running these ChatGPT-style searches en masse, said University of Washington professor Luis Ceze.
"Search is arguably the world’s most ubiquitous technology," he told Yahoo Finance. "For context, Google processes around 8.5 billion searches per day... Because generative models are exceptionally large and complex, this process is computationally intensive, making it far more expensive to run than other kinds of deep learning models."
In the end, the high expectations the market has for Google here are bumping into a reality that no one totally expected, said Michael King, founder and CEO of SEO agency iPullRank.
"We wanted to see something that was a game-changer and instead what we ended up with was them talking about subtleties," King said. "We were expecting to be amazed by this new integration and instead, we got 'did you know we invented the technology?'"
"This is really feeling like a role reversal," said King.