Hello, fellow humans! Hope you’re enjoying the new, limited Saturday edition of the Daily Brief, which is focused on AI, though, we promise, written by actual people.
Got some questions about AI you’d like answered? Or just some AI hallucinations you’d like to share? Email us anytime. Enjoy!
Here’s what you need to know
Baidu said it’s readying major updates to its Ernie bot. The Chinese technology company teased the upgrades in an ad in which Ernie goes missing and people forget how to function.
US senators drafted legislation that would protect performers from AI impersonations. That generated Drake and Weeknd song must have really jarred lawmakers enough to introduce the No Fakes Act.
AMD, one of the world’s largest computer chip firms, acquired an AI startup. AMD wants to shore up competition with its rival Nvidia; the acquisition comes as generative AI deals are trending down (more on that below).
And in another downturn, AI startups are laying off workers. Founders are placing the blame on high interest rates, which have translated to less funding.
OpenAI is close to signing the biggest office deal in San Francisco since 2018. AI is boosting the city’s office glut.
AI deals are already on the decline, at least for now
Deals for generative AI were down 29% in the third quarter of this year compared to the one before it, with their values totaling $6.1 billion—an amount that was mostly made up by Amazon’s $4 billion agreement with language model provider Anthropic.
There are a couple factors that could be leading investors to pump the brakes. One is that Big Tech, with its market influence and deep pockets, might be scaring away smaller players. But even Big Tech has slowed down its dealmaking. And valuations are still all over the map, further complicating things.
Despite these shifts, the AI hype isn’t dying—rather, investors are narrowing their bets, and they’re hyper-focused on companies with enterprise AI that can lower back office costs and take care of boring stuff.
Pop quiz: Around the world in AI regulation
Which country has drafted rules that would require providers of AI systems to submit a risk assessment before rolling out the product to the public?
A. Japan
B. Brazil
C. US
D. Italy
We’ve got the answer right here (along with what other countries are doing), or you can scroll down to the bottom of this email to find out.
Quotable: The harsh reality of AI watermarking
“We broke all of them.”—University of Maryland computer science professor Soheil Feizi in an interview with Wired about the current use of watermarking in AI images, even as the G-7 wants to ask companies to use them.
Other great AI reads
🪧 Striking actors aren’t closer to a deal that would address their AI concerns
🥱 Canva releases AI tools to automate boring design tasks
🏃 Zoom hopes its newest AI features will give it an edge over Microsoft Teams
🖱️ The difference between building, buying, and renting AI talent
🤖 SoftBank’s CEO says AI will surpass human intelligence in a decade
🤳 Google’s latest Pixel phones are packed with AI
Ask an AI—or a group of them at once
Character.AI lets you chat with avatars of famous personalities (or just, well, personalities—think “psychologist” or “English teacher”). They’re all user-generated, in that you tell the tool what your character is like and add example messages.
To date, Character.AI has raised $190 million in funding from investors like Andreessen Horowitz and Greycroft. And though the company says users spend an average of two hours on the site per day, it’s easy to see it getting old pretty fast. That’s where Character.AI’s new premium feature comes in: A group chat of up to five AI chatbots and five other humans at once.
We decided to assemble some economists—say, John Maynard Keynes, Adam Smith, and Janet Yellen—next to someone having a big effect on the global economy right now: Taylor Swift. We also added a talking sloth, for fun.
But though we did get some broad-strokes financial analysis from AI Yellen, our economists mostly just wanted to keep talking about how great Taylor was. The sloth really wanted to talk about leaves.
The more we talked, the more everyone started to sound alike, even using the same phrases over and over, so we swapped out Keynes for William Shakespeare. While AI-speare did sound pretty erudite, he didn’t exactly write like the Bard. Immediately, everyone changed their tone to match his, even the sloth.
Could this feature ever have useful applications other than playing around on a Friday afternoon? If the technology becomes more sophisticated, having Adam Smith discuss economics with Janet Yellen would be enlightening—as would being able to make a couple of characters from your favorite TV show work out their differences in a way that you wish they’d done back in season 2. But it’s not there yet.
Our best wishes for a very human day. Send any news, comments, chatbot transcripts, and leaves to [email protected]. Reader support makes Quartz available to all—become a member. Today’s AI Daily Brief was brought to you by Michelle Cheng, Morgan Haefner, Gabriela Riccardi, and Susan Howson.
The correct answer to the pop quiz is B., Brazil. Regardless of the risk classification of AI systems, regulators say that people affected by those systems have the right to an explanation about a decision or a recommendation made within 15 days of the request. Read more about what other countries are saying about AI regulation.