Artificial intelligence is more a part of our lives than ever before. While some might call it hype and compare it to NFTs or 3D TVs, AI is causing a sea change in nearly every facet of life that technology touches. Bing wants to know you intimately, Bard wants to reduce websites to easy-to-read cards, and ChatGPT has infiltrated nearly every part of our lives. At The Verge, we’re exploring all the good AI is enabling and all the bad it’s bringing along.
Featured stories
Q&A: Mark Zuckerberg on winning the AI race
Meta’s CEO is pushing for the company’s assistant to be the most used chatbot in the world. He opens up about competing with other AI companies, model training, and why he bought all those GPUs when he did.
If it pushes through, the European Commission could look into whether Microsoft’s $13 billion investment into OpenAI restricts competition, Reuters reports. The commission floated the possibility of an investigation in January.
The partnership between the two companies already drew scrutiny from the UK’s Competition and Markets Authority and the Federal Trade Commission in the US.
As we saw in the beta release this new version of Adobe Express packs the same creative, editing, and generative AI features that desktop users have into an iOS and Android app.
It’s free to use, but to access Firefly and the full suite of editing tools you’ll need a $10 per month Premium membership.
A group of key senators introduced the Future of AI Innovation Act Thursday to promote artificial intelligence research and innovation and create global standards for the technology.
The group’s sponsors include Commerce Committee Chair Maria Cantwell (D-WA), Todd Young (R-IN), a member of Majority Leader Chuck Schumer’s AI working group, and Marsha Blackburn (R-TN), and John Hickenlooper (D-CO).
[www.commerce.senate.gov]
From ChatGPT to Gemini: how AI is rewriting the internet
How we use the internet is changing fast thanks to the advancement of AI-powered chatbots that can find information and redeliver it as a simple conversation.
Alex told us they were coming, but it looks like Meta is flipping the switch as we speak — adding Meta AI searches alongside traditional ones in each app’s default search box. Oh, and some Ask Meta AI mini-prompts, too. If Meta AI has an answer to something you type, it’ll show a blue-violet circle next to that search result.
At Kernel, your veggie burger will be served by a robot
Its robotic arm heats vegan burgers and crispy potatoes while relegating humans to assembly line jobs.
ChatGPT is winning the future — but what future is that?
OpenAI didn’t mean to kickstart a generational shift in the technology industry. But it did. Now all we have to decide is where to go from here.
Autonomous Cars
Orange cones are robot kryptonite.
Senate Democrats urge crackdown on autonomous vehicles and driver assist
People are afraid of self-driving cars — can the industry change that?
Tesla slashes price for monthly Full Self-Driving subscription
The deepfaked voice of Anthony Bourdain in Roadrunner is just one controversial application of AI in documentaries, and now the Archival Producers Alliance has drafted some “best practices.”
It’s calling for filmmakers to be “transparent” to viewers, to get consent from human subjects, and to prioritize primary sources. The group hopes to issue final guidelines this summer.
[The Hollywood Reporter]
404 Media reports that a parent soliciting advice in a Facebook group got something unexpected when Meta’s AI chatbot chimed in with some thoughts. The reply starts with “I have a child who is also 2e,” which refers to a child who is both academically advanced and has a disability. Let me stop you right there, AI chatbot.
A small, sparkly ghost logo will “soon” feature on saved or exported images created with Snapchat’s generative AI tools.
More information — including some fairly obvious yet rigorous generative AI dos and don’ts — can be found on Snap’s support page.
Inside Google’s big AI shuffle — and how it plans to stay competitive, with Google DeepMind CEO Demis Hassabis
Google invented a lot of core AI technology, and now the company’s turning to Demis to get back in front of the AI race for AI breakthroughs.
The investment into G42 brings Microsoft’s AI services and chips to the United Arab Emirates. Meanwhile, G42 will run on Azure. Microsoft president Brad Smith will also join G42’s board.
The New York Times reports the Biden administration helped broker the deal. Under the transaction, G42 will agree to remove Chinese technology from its operations.
[The New York Times]
That’s the question being raised by its Oversight Board, which today announced two new cases looking into how Meta handled explicit fakes of female public figures posted to Facebook and Instagram. One of which could concern the fake Taylor Swift images that circulated online earlier this year.
The board’s investigation will take a few weeks before reaching a final non-binding decision.
OpenAI’s Batch API now lets users upload a file of bulk queries to the AI model, like categorizing data or tagging images, with the understanding that they won’t need immediate attention. Promising results within 24 hours lets them run when there is unused compute power, and keeps those pricey GPUs humming around the clock.
I often mention AI model benchmarks in posts, but Kevin Roose at The New York Times said the quiet part out loud: AI benchmark tests don’t help in comparing models, and these need to change.
Benchmarks cover a small amount of human knowledge, but as Roose points out, AI models easily surpass that. Training datasets sometimes include answers from benchmarking tests, so, of course, models beat the tests.
[The New York Times]
A bunch of places covered AI-generated images of an unholy Jesus/shrimp hybrid going viral on Facebook earlier this year, but the attention didn’t cause Zuck to take any action to slow the situation down. Here’s JC Noods laying back on a Lambo, a post which has 36,000 likes on Facebook right now. The AI internet is going great, y’all.
The new deal adds protections around using digital voice replicas, reports Reuters.
The terms “artist,” “singer,” and “royalty artist,” under this agreement only include humans. In this agreement, clear and conspicuous consent, along with minimum compensation requirements and specific details of intended use, are required prior to the release of a sound recording that uses a digital replication of an artist’s voice.
AI voice cloning improvements mean we’ll probably see more agreements ensuring performers are informed about how copies of their voices are used.
If you thought David Pierce’s review for us was scathing, just wait till you watch Marques Brownlee’s review that just dropped. Yikes.
Investor funding for AI startups fell for a second year, according to Stanford. Sounds like the reckoning is here!
Several things could be in play: recognition of AI’s challenges, less spending on expensive foundation models, and “the rapid rise and fall of certain marquee name startups in AI.” 👀
OpenAI chose Tokyo for its first office in Asia as it expands its footprint outside of the US. The company is also releasing a version of GPT-4 in Japanese.