What happens when you miss two weeks of AI news | The AI Beat

What happens when you miss two weeks of AI news?

It’s simple: You are left in the dust of a news cycle so fast-paced that even ChatGPT would likely respond that “as an AI language model, I am unable to update you on such a deluge of AI news.”

That’s what happened to me when I returned from a long-awaited overseas family vacation this past Friday.

Of course, I knew that taking so much time away from the AI news cycle, with news apps and social media removed from my phone, was dicey. After all, since I started at VentureBeat in April 2022, shortly after OpenAI’s DALL-E 2 was released, the pace of AI development and news-making has steadily sped up. And the months since ChatGPT became a household name nearly overnight have felt like riding a Japanese bullet train that doesn’t pause long enough for anyone to get off.

But my vacation offered the opportunity for a welcome and sorely-needed break from AI Twitter beefs and the usual breathless AI headlines. And while I admit that I took a quick peek towards the end of my trip, for the most part I stuck to my guns and avoided the daily AI-news adrenaline surge I’ve gotten used to over the past year. Sifting through the digital pile of AI stories when I returned, however, was both overwhelming and fascinating.

Not on my bingo card: Geoffrey Hinton warning of AI dangers

One major story not on my 2023 bingo card was Geoffrey Hinton’s May 1 announcement that he had quit his job at Google so he can freely speak out about the risks of AI. According to the New York Times, Hinton said a part of him now regrets his life’s work.

It was quite the turnaround from when I spoke to Hinton last August, for a story on the 10th anniversary of the seminal AlexNet paper that jump-started the subsequent deep learning “revolution.” Back then, he was jubilant about the recent progress of LLMs.

“We’re seeing these amazing things that the big language models can do, which we thought would be a long way off,” he told me back then. He also added: “Any time you’ve got a lot of data, and you want to make predictions from it, then deep learning is going to be extremely helpful. That’s why all the big companies like Apple and Google and Microsoft and Facebook, and Amazon…rely a lot on deep learning. It’s going to be relevant to more or less everything.”

Biden Administration announces efforts to tackle AI risks

I wasn’t surprised that Vice President Kamala Harris and other senior Biden Administration officials met with the CEOs of Alphabet, Anthropic, Microsoft and OpenAI on May 4.

I had already reported on the Blueprint for an AI Bill of Rights back in October as well as the January release of the NIST AI Risk Management Framework, and the meeting also came in the context of recent moves in AI regulation. Most notably, there was the draft of the highly-anticipated EU AI Act, which was passed on April 27 and got the votes it needed last week to move ahead to the next stage of the legislative process.

And the fact that so many new initiatives were announced — including $140 million in funding to launch seven new National AI Research Institutes, public assessments of generative AI systems, and draft policy guidance on the use of AI systems by the U.S. government, shows that the Biden Administration knows it needs to get serious. But can they go beyond glittery PR to move quickly as the breakneck pace of AI development continues?

Google goes big and OpenAI comes out swinging

What a change since Google’s limp Bard debut in Paris back in February! Back then, the company’s muted response to Microsoft’s search “race” with Bing had many predicting that Google had missed its moment in generative AI, even though it was Google’s Transformer neural network architecture, which launched in 2017, that had made today’s generative AI gold rush possible.

Last week’s Google I/O changed all that: According to CNBC reporting this weekend, even Google employees gave credit to engineers for their “quick work” on AI products, while poking fun at the fact that executives said “AI” over 140 times in the Google I/O keynote. Among other announcements, Google announced Bard upgrades in a bid to challenge ChatGPT, debuted PaLM 2, a next-generation language model, and released a slew of new generative AI features in Google Cloud and Vertex.

But don’t sleep on OpenAI — the company was not to be outdone last week. Right after I landed at JFK Airport on Friday, I got an email from the company’s PR representative, letting me know that ChatGPT plugins will be available in beta this week to ChatGPT Plus users. The generative AI race (yes, it is a race) continues.

Gearing up for another epic AI news week

Still jet-lagged, I was up early this morning sifting through two weeks’ worth of emails and gearing up for another epic AI news week. First on my list: OpenAI CEO Sam Altman’s congressional testimony tomorrow.

Anything I should know as I get back into the post-vacation swing of things? Feel free to DM or email (Twitter: @sharongoldman)

Okay…let’s go! (Rocky theme music plays in the background)