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 part of the technology industry. OpenAI’s ChatGPT is arguably the best-known AI chatbot around, but with Google pushing Gemini, Microsoft building Copilot, and Apple working to make Siri good, AI is probably going to be in the spotlight for a very long time. At The Verge, we’re exploring what might be possible with AI — and a lot of the bad stuff AI does, too.
Featured stories
Filed under:
AI is confusing — here’s your cheat sheet
If you can’t tell the difference between AGI and RAG, don’t worry! We’re here for you.
Henry Farrell, co-author of Underground Empire: How America Weaponized the World Economy, notices that LLMs do an awful lot of things done by middle management. What if the wrong people are worried that AI will take their jobs?
[www.programmablemutter.com]
The “Grok 2 + Aurora” option has vanished from Grok’s model selector menu only a day after it appeared, Engadget reports, replaced by “Grok 2 + FLUX beta” instead. The model still makes photorealistic images, but it was less willing to reproduce celebrities when I asked.
X owner Elon Musk wrote yesterday that the photorealistic and largely unrestricted model is a beta “internal image generator.”
Update: Added testing detail.
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.
Google’s AI weather prediction model is pretty darn good
The company says its AI model outperformed a traditional forecasting system.
What Arm’s CEO makes of the Intel debacle
Why Intel’s longtime ‘frenemy’ thinks the chipmaker getting out of its mess ‘may be too big of a hill to climb.’ Plus: the scoop on Samsung’s leadership shuffle, Bezos kisses the ring, and Altman lowers the AGI bar.
OpenAI just announced an alpha program for a new tool (called reinforcement fine-tuning) that lets developers train models on specific tasks, using example problems and answers. In a post after the livestream announcement, CEO Sam Altman said this will make it “really easy to create expert models in specific domains with very little training data.”
From Perplexity’s blog post:
Today, we’re excited to welcome over a dozen new partners to Perplexity’s Publishers’ Program: ADWEEK, Blavity, DPReview, Gear Patrol, The Independent, Lee Enterprises, Los Angeles Times, MediaLab, Mexico News Daily, Minkabu Infonoid, NewsPicks, Prisa Media, RTL Germany brands stern and ntv, and World History Encyclopedia.
Perplexity kicked off the program earlier this year by signing deals with Time, Der Spiegel, Fortune, and others.
[www.perplexity.ai]
Autonomous Cars
Uber’s first international robotaxi service is live in the UAE
Waymo’s next robotaxi city will be Miami
GM wants to protect its robotaxis from ‘adversarial’ pedestrians.
The small thing that can keep drivers attentive while using partial automation
A new Llama 3.3 70B model dropped today, and Meta’s VP of genAI said it matches the performance of the larger 405B version, but runs cheaper and lighter. They also claimed this new model beats competitors Google, OpenAI, and Amazon on key tests like MMLU.
Competitors also really love overlapping their AI announcements — Google, Microsoft, OpenAI, and xAI have all dropped announcements this week.
Microsoft is stepping up its Copilot Plus PC marketing today with a new ad for the upcoming Captain America: Brave New World movie. The 60-second spot features Danny Ramirez’s Joaquin Torres (Falcon) using a Surface Laptop 7 with Paint’s Cocreator AI mode to generate images. Microsoft has been running similar TV ads in recent weeks, claiming that its Qualcomm-powered laptops are somehow “the fastest... Windows PCs ever.”
AGI is coming and nobody cares
On The Vergecast: the goalpost-moving AI industry, cable is once again so back, and a bit on our new subscription.
Stop using generative AI as a search engine
A fake presidential pardon explains why you can’t trust robots with the news.
The people suing Minnesota over its law reining in political deepfakes have submitted a filing (PDF) asking the judge to withdraw a declaration supporting the law.
Why? Because, as Stanford misinformation expert Jeff Hancock recently admitted, it was written with help from ChatGPT, which generated incorrect citations in the document.
Frank Bednarz, the lawyer for the defendants, tells The Verge:
Minnesota Attorney General Keith Ellison’s decision not to retract a report they’ve acknowledged contains fabrications seems problematic given the professional ethical obligation attorneys have to the court.
[hlli.org]
