Google’s I/O conference will be a key test of its AI strategy

On Wednesday, Google is holding its annual conference to showcase its latest innovations — a 15 year-old ritual that this time around is upset by the tech giant’s suddenly playing catch-up in a field it had long dominated: artificial intelligence.

The explosion of hype around new breakthroughs in AI tech has thrown Google onto the defensive, forcing it to try to regain the initiative after its archrival Microsoft and smaller companies like OpenAI put new “generative” AI tools directly in customers’ hands. Google’s event, known as I/O — short for input/output, a software term for human-computer interaction — is a chance for executives to show skeptical investors, competitors and, in many cases, their own employees, that it is still the leader in AI.

The buildup to the event has added to the intense environment Google workers have found themselves in after their bosses sent out orders to redouble their efforts on AI in December, after OpenAI’s ChatGPT chatbot became an instant phenomenon.

The expectation that Google stage a blowout AI event is tempered by the current frugal mood at the company after it laid off 12,000 employees in January and cut some of its famous perks. The conference is much smaller than in years past: What used to be a three-day extravaganza covering huge swaths of the company’s Mountain View, Calif., campus is now just one day. This moment is the most stressful that workers can remember at the company, according to conversations with five current and former employees who spoke on the condition of anonymity for fear of retribution.

“It’s a different time at Google,” one current employee said. “There’s a lot of pressure now.”

Generative AI’s rival factions

Showing off new tech to customers, the media and investors is key given the perception from analysts and industry observers that Google fumbled its March launch of the “Bard” chatbot, four months after OpenAI debuted ChatGPT and after Microsoft rebooted its Bing search engine with ChatGPT. For most of its two decades, Google has enjoyed a reputation as the undisputed leader in its core business areas. Google Search has no serious competitors, and Google Maps, Gmail, and the Chrome web browser dominate their product categories so deeply that antitrust authorities in multiple countries have launched investigations or filed lawsuits alleging that the company is breaking competition laws. That dominance allowed the company to grow ever bigger, hiring thousands of new employees in the past few years and expanding into new product areas.

But the advent of generative AI models, which have ingested trillions of words and phrases from the World Wide Web and can now have eerily humanlike conversations, has prompted some AI leaders and tech pundits to suggest that internet searches will be replaced by AI bots that can directly answer specific questions.

“Microsoft and OpenAI right now are the companies that all others need to keep pace with” said Fred Havemeyer, an analyst with Macquarie Group. “Google is playing catch-up.”

Google spokesman Chris Pappas declined to comment.

I/O is the company’s biggest annual event, and in the months and weeks leading up to it executives and product managers debate what exactly to focus on. As many as a quarter of the company’s nearly 200,000 employees may be involved in something that will be announced or showcased at I/O, said one former employee.

Whimsical dome-like structures are set up across acres of parking lots at the company’s campus, each filled with colorful games, hands-on demonstrations and brand-new hardware. But this year, far fewer people have been invited to attend in-person, and most developers and partners will be streaming remotely instead of flying to California, something that for many developers who work in the Google ecosystem was a coveted once-in-a-lifetime opportunity.

Inside the secret data sources that make chatbots like ChatGPT sound smart

At I/O, the company is expected to finally detail how it will begin using generative AI in search results. Google’s leaders have been emphatic that Bard is not meant as a replacement for search.

But the company has already spoken about how generative AI can be helpful when it comes to answering complex queries with no single answer, something that is currently done by bloggers and content creators online. Google could also use a search chatbot to look through different answers and summarize them, getting rid of the need for people to click through multiple links when looking for information. Either way, overhauling how search results are presented could upend the years-long relationship the company has with internet content creators, who often need traffic from Google search to generate revenue.

Google has called itself an “AI first” company for years, and tech breakthroughs made by its researchers have helped seed the entire AI ecosystem. Its own employees invented transformers, a type of AI model that could digest larger data sets, and after publishing their findings, saw the tech taken up by companies that have now become key competitors. It has threaded AI into existing products, such as improving the auto-complete that shows up in the Google search bar, or vastly increasing the quality of translations done by Google Translate. The company developed its own AI chatbots, and even showed off its LaMDA chatbot at the 2021 I/O event.

But Google hesitated to make the new AI tools broadly available. Chatbots have a penchant for making up false information. The initial blog post announcing Bard unintentionally included an error the chatbot had made. A March event detailing some of the company’s generative AI plans was written off by investors as underwhelming, leading to the company’s value falling by $100 billion in a single day.

The event will illuminate the question of just how fast Google is willing to move to answer the competitive threat from Microsoft and OpenAI. Some employees have cheered the newfound urgency and argue that getting its tech into people’s hands while it is still in an experimental state is important to help it keep up with smaller, more nimble competitors. Other employees, as well as outside ethical AI researchers, worry that the new tactics could lead to products that can cause harm getting into the hands of regular people.

“This is a tenuous moment for Google in the AI arms race,” said Dan Ives, a stock analyst with Wedbush Securities. “This is Google’s time to restart the AI engines with time ticking.”

Add a Comment

Your email address will not be published. Required fields are marked *