To the surprise of nobody (except maybe the delusional mophead himself), Sam Bankman-Fried is in custody on numerous criminal charges and facing a lawsuit from the Securities and Exchange Commission. The fine folks at Fortune Crypto are on the case, and you can follow the latest developments here.
In the meantime, let’s take one last look at generative A.I.—and in particular, Big Tech’s absence from the craze—before the year comes to a close. (An early reminder: Our last Data Sheet of 2022 will be published on Friday, and we’ll return on Jan. 4.)
When the tech history books are written on 2022, generative A.I. will occupy an early chapter in the tome.
First came the emergence of text-to-image generators DALL-E 2 and Stable Diffusion, opening our eyes to the possibilities of consumer-facing A.I. Then came the arrival of ChatGPT, a stunningly intelligent and creative chatbot. Finally, the A.I.-powered photo and video editor Lensa burst onto the scene, with TechCrunch reporting Monday that it’s currently the world’s most-downloaded app.
The technology alone is remarkable enough. But one of the underappreciated parts of generative A.I.’s coming-out party has been its hosts.
Despite billions of dollars at their disposal, the tech industry’s biggest behemoths have essentially ceded the early buzz over generative A.I. to companies barely known outside of Silicon Valley.
DALL-E and ChatGPT are the products of OpenAI, a six-year-old research lab-slash-startup (admittedly, one bankrolled with more than $2 billion by Microsoft and a roster of tech luminaries). Stable Diffusion developer Stability AI just raised its first $100 million funding round. And Lensa creator Prisma Labs has only announced a single funding round, totaling $6 million in 2019, per Crunchbase.
The relative anonymity of these companies raises the question: Did Big Tech, weighed down by its bloat and complacency, miss the generative A.I. bash?
Well, not exactly. Compared to the scrappy startups nipping at their heels, tech industry giants face a more complex calculus when rolling out novel technology.
Take, for example, Alphabet. The Google parent has shoveled countless dollars into developing LaMDA, a large language model akin to ChatGPT, yet company officials haven’t released their chatbot to the public yet.
In many ways, this approach is understandable. Alphabet, already a known commodity, doesn’t benefit from buzz in the same way as OpenAI. Rather, its priorities are financial, which means integrating A.I. technology into its existing products—a process that could take years to complete—and doing so in a way that maintains consumer trust in the company.
“How Google missed this moment is not a simple matter of a blind spot,” Alex Kantrowitz, author of the Big Technology newsletter, wrote last week after speaking with an early LaMDA product manager. “It’s a case of an incumbent being so careful about its business, reputation, and customer relationships that it refused to release similar, more powerful tech.”
In a similar vein, major tech companies arrive at the generative A.I. scene carrying much more baggage than upstarts, putting added pressure on them to deliver safe and effective products. While users have marveled at the magic of image generators and chatbots, they’ve also spotted major problems with the technology. The issues range from proliferation of false information, to the creation of deepfake videos, to potential violations of artist copyrights.
Google senior vice president James Manyika spoke of these considerations in a discussion with Fortune CEO Alan Murray last week at the Brainstorm A.I. conference in San Francisco: “Google has been extraordinarily thoughtful about not rushing to throw these things out into the world without actually doing proper research and fully understanding what these technologies are capable of.”
It’s a lesson that Meta, hardly a paragon of public trust, learned the hard way earlier this year.
The Facebook parent made a terrible first impression this summer when it debuted Blenderbot, the company’s new A.I. chatbot. The product looks rather rinky-dink compared to ChatGPT, and users immediately prodded it into spewing racist comments and conspiracy theories. While Meta could eventually take the lead in generative A.I., company executives did the firm no favors by releasing something as underwhelming as Blenderbot.
Fair questions can be asked about whether some larger tech outfits would have benefited from getting farther ahead of the generative A.I. curve. Lensa’s ascendance, for example, sure feels like a missed opportunity for Snap, which has seen its stock price plummet 80% this year.
But as we head into 2023 and beyond, most members of Big Tech will benefit more from arriving at the generative A.I. party fashionably late, rather than getting there first.
Want to send thoughts or suggestions to Data Sheet? Drop me a line here.
Jacob Carpenter
