Silicon Valley’s AI Frenzy: A Bubble Full of Problems, Doomed to Burst?

The OpenAI logo is displayed on a cellphone with an image on a computer screen generated by ChatGPT’s Dall-E text-to-image model, Dec. 8, 2023, in Boston. European Union lawmakers were set to give final approval Wednesday to the 27-nation bloc’s artificial intelligence law, setting up a sweeping set of rules for the fast-developing technology that is expected to take effect later this year. (AP Photo/Michael Dwyer, File)

Huge sums are being poured into generative artificial intelligence, a technology that originated in the San Francisco Bay Area and has taken the world by storm. But the relentless hype is inflating a bubble that experts say is doomed to burst. So far, the return on investment has been modest, while serious problems are multiplying.

It’s been less than two years since San Francisco’s OpenAI launched its generative AI bot ChatGPT, sparking a Big Tech arms race, a torrent of venture capital funding for AI startups, and a bandwagon of companies looking to cut costs and boost productivity by embedding the technology into virtually every product and service imaginable.

Investors have poured more than $24 billion into generative AI, according to consulting giant EY, and tech companies plan to spend $1 trillion on AI infrastructure in the coming years, Goldman Sachs predicts. Many technologists see huge potential in the technology that uses patterns and relationships in data to generate text, images, and audio, while others see critical gaps.

“Everybody wants to make money in the AI ​​race,” said Howard Young of San Jose, who is integrating AI software into the computing systems of tech giant AAEON to improve urban infrastructure, industrial processes, manufacturing and medicine. Young joined hundreds of other tech workers and executives this month at the Reuters Momentum AI conference in San Jose. “The real organic revenue, I don’t see it yet,” Young said. “It’s going to take time, even with the best minds in Silicon Valley.”

Tech industry players “significantly overestimate” the current capabilities of generative AI, and the extent of its improvement remains uncertain, said Jim Covello, chief global equity researcher at Goldman Sachs.

“The technology is nowhere near where it needs to be to be useful,” Covello said in a bank newsletter focused on generative AI in June. “If AI technology ends up having fewer use cases and lower adoption than the current consensus expects, it’s hard to imagine that won’t be a problem for many companies spending on the technology today.”

David Cahn, a partner at Silicon Valley venture capital giant Sequoia, used a June blog post to highlight a “speculative frenzy” around generative AI leading to a “delusion” spreading out of Silicon Valley “that we’re all going to get rich quick.”

Generative artificial intelligence, dubbed “genAI” in tech circles, is creating a bubble that’s bound to burst, and damage is on the horizon — but it’s nothing new to Silicon Valley, said Steve Blank, an assistant professor of management science and engineering at Stanford University. Blank compared the technology to the meteoric rise of the World Wide Web and the dot-com crash that followed.

“It’s not that we were wrong about the web, it just took several iterations and some reworking to separate the wheat from the chaff,” Blank said.

For Bay Area startups, it’s not far from the truth to say, “You’re not going to get funding if you don’t have AI in your title or your story,” Blank said. “It’s insane. That giant sucking noise you hear is all these lemmings throwing money at what’s the next big thing. One or two of them will strike gold. The rest of them will lose their shirts.”

The inevitable bursting of that bubble may not be as damaging as the implosion of the dot-com bubble, “simply because many of the companies that are spending money today are better capitalized than the companies that were spending money back then,” said Goldman Sachs’ Covello.

The business transformation touted in support of massive spending on generative AI has failed to materialize, with the expensive technology largely incapable of solving the complex problems that would enable widespread automation of tasks and jobs.

In the meantime, the technology remains plagued by problems that are, depending on who is talking about it, either growing pains or fundamental flaws. Leading developers of generative AI are battling artists, photographers, authors, coders, music labels and newspapers — including this one — in court over alleged theft of copyrighted material by scraping the internet to “train” AI models.

Training and leveraging generative AI has upended Google and Microsoft’s progress toward their climate and sustainability goals, with both companies reporting dramatic increases in electricity and water consumption last year due to AI-related data processing and storage. Chatbots and generative search continue to produce errors and lies. Propagandists use the technology to spread disinformation, students use it to cheat, and criminals use it to scam and harass. State legislatures introduced nearly 200 bills last year to oversee and regulate AI.

Yet those who believe in the promise of generative AI see innovation overcoming most challenges, and they point to important early uses and powerful potential applications.

“The industry is just forming,” Blank said. “The earth is still molten. We’re starting to see the outlines of the continents.”