Yesterday, Google's annual I/O developer conference kicked off under Shoreline Amphitheater's large circus-like ceiling, with improviser and YouTuber Marc Rebillet donning a rainbow-colored robe and holding a giant Google coffee. I climbed out of the cup. He took to the DJ booth to show off Google's new generative AI music tool called MusicFX. The demo was filled with pounding beats meant to wake up “stupid nerds,” he exclaimed.
After a few days of serious build-up leading up to this week's AI announcements, it turned out to be a shockingly silly moment that included a pre-emptive product launch from OpenAI. Social media has skyrocketed expectations about how OpenAI will steal Google's thunder. Announcement of unique ChatGPT search? Surprise rollout of GPT-5?
For a moment, it was easy to imagine what Google would become. screwed Due to OpenAI's proactive measures. More precisely, I unscrewed it again. After all, a year ago there were a lot of headlines like “AI first” for the last time: How Google has fallen behind in the AI boom. After ChatGPT was launched in November 2022, there were reports that Pichai declared a “code red” internally in response to it, and he spoke out about the threat that OpenAI's partnership with Microsoft poses to Google's business. became very panicked. OpenAI because Microsoft was “very concerned” that Google was far ahead in AI, at least from the outside looking in, given that Google was far ahead in AI in 2019. This seemed like a humiliating acquisition given the company's decision to invest in . And at the 2023 Google I/O conference, the company appeared decidedly defensive, repeatedly reminding the audience that Google was the first “AI-first” company, and incoherently and poorly thought out. A large number of product announcements and demos that have not been missed. The whole thing was rushed and seemed like Microsoft and he were simply designed to match OpenAI.
This year it was a completely different atmosphere. As of yesterday afternoon, both OpenAI and Google had finished their live streaming announcements. All that remained was the memory of OpenAI's emotional voice assistant. Google's evolution of AI overview in search. It was clear why Pichai seemed more relaxed this year, including several demos featuring dogs. That's because far from messing with AI, Google is actually pretty calm about it. Not only has the company caught up to the point where its Gemini model is considered top-of-the-line, but it also has the strength of a moat that OpenAI cannot easily cross: Google's vast consumer distribution across a deep and broad product mix. I have a newfound confidence in my strength. It has billions of users. Many have accused Google of being bloated, bureaucratic, and slow.But it's unparalleled scale.
Jim Fan, senior research manager and head of embodied AI at Nvidia, put it succinctly in yesterday's X post: “Google's strongest moat is distribution,” he wrote. “Gemini doesn't have to be the most used model in the world.”
OpenAI may have a technological lead. GPT-4 continues to be the most advanced AI model that can excel at certain inference tasks. And as a startup with an $80 billion private market valuation, you might be able to: It works faster than Google. But Google has Search, Gmail, Maps, Photos, Calendar, Drive, Docs, Sheets, and Android. By integrating generative AI into all of these products and using AI to tie them all together (for example, an AI-powered workflow that scours Google applications for data to pull into Sheets or Calendar), you can continuously engage your users. You can continue to use it. . Even the Google search box is already working with AI across a variety of products, offering options such as finding photos, translating text, solving homework with Google Lens, identifying songs, and buying products with screenshots. I am.
And if Google's distribution is its most powerful moat, its secret sauce is arguably the sheer volume of petabytes flowing through its products every day from web searches, YouTube videos, map requests, customer reviews, and more. This is the data. In addition to training models, users can have Google's AI capabilities examine data stored in Google apps to assist with an ever-increasing number of personalized tasks. So you can basically use Google to search your own Google.
Of course, this doesn't mean Google's future AI success is guaranteed. While its rich product portfolio and broad reach may help it grow its user base, when it comes to revenue, the company's business model is focused on search. It also remains to be seen whether the rollout of AI overview in search will cannibalize referral and advertising business. And how badly AI-generated content pollutes the entire information ecosystem, resulting in poor search results and dissipating users, whether for AI-generated or ranked links. It remains unclear whether they will be moved away from each other. Additionally, while it cannot be as nimble and nimble as OpenAI, Google needs to avoid embroiling itself in “blame” battles, as it did during the “woke” image controversy over Gemini in February. there is.
However, OpenAI is likely to face an uphill battle. If you want to beat Google with consumer AI products, you need to show that you're deploying significantly better AI models than Google, and you need to keep deploying those AI models to stay ahead. There is a need. A rumored partnership with Apple to add ChatGPT to the iPhone might help. Apple, despite its partnership with Microsoft, has a distribution lacking in OpenAI. But Google has decided to move away from the AI that most consumers would use without announcing that the company has somehow reached AGI (what it thinks AGI is). proved to be extremely difficult.
