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, generative AI is causing a sea change in nearly every part of the technology industry. OpenAI’s ChatGPT is still the best-known AI chatbot around, but with Google pushing Gemini, Microsoft building Copilot, and Apple adding its Intelligence to Siri, 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.
Neon is three AI tools dressed up as one, with a $20-per-month subscription attached.
US power grids aren’t moving fast enough to keep up with the sudden rise in electricity demand from AI. Data center developers are forging ahead anyway, adding their own gas turbines and fuel cells.
[The Wall Street Journal]
Plus a sore neck, a peeved spouse, and ethical quandaries.
Microsoft thinks the future of Windows involves letting AI take charge. It might be right, but only if you’re willing to embrace a little risk.
Sophisticated_Bean_3:
Tired: the certainty of clicking a button or entering a command
Wired: the uncertainty of delegating mundane tasks to a token predictor
Get the day’s best comment and more in my free newsletter, The Verge Daily.
At least, that’s what plummeting human pageviews suggest. In the last few months, it’s down 8 percent from last year after implementing improved bot detection measures.
Wikipedia is hardly alone; AI keeps users away while using its content for answers. Even if Wikipedia manages to survive, difficult times are ahead.
Remember how Salesforce billionaire Marc Benioff suddenly looooved Donald Trump? It turns out Salesforce is trying to sell an AI product to ICE. Two things: 1. This is how the gangster tech era works. 2. If I were a betting woman, I’d put money down that Agentforce is floundering. Certainly it’s been expensive marketing it.
[The New York Times]
Spitfire has a good check-in on the inevitable collision of hyper-documented influencers and hyper-realistic AI video:
Even if people don’t believe or trust this AI-generated video content of influencers and other public figures, some people will, because a lot of people already really struggle to identify when media is AI-generated and it’s constantly getting harder to tell. And even fake media can be part of a bullying or smear campaign.
[Spitfire News]
Photos of the UAE datacenter construction site released Thursday show the footprints of the big buildings that OpenAI and its Stargate partners, Oracle, NVIDIA, Cisco, and SoftBank say will house 16-gigawatts of compute power for everything from curing cancer to sexting. Thus far, construction of the first 200 megawatts is “well underway” to meet the 2026 deadline.
Bloomberg reports that Ke Yang, only just promoted to head of the team, is leaving for Meta. He was tasked with building a Siri search tool to rival ChatGPT and Gemini, and is one of a dozen or so prominent AI departures for Apple this year.
The company says it codes just as well as Claude Sonnet 4 but does so “significantly faster” and at a third of the cost. Anthropic’s advanced model is the recently released Sonnet 4.5, which can “break down a complex problem into multi-step plans” and manage several Haiku 4.5s to implement the subtasks concurrently. Haiku 4.5 is now available on free Claude plans.
Update: Free Claude plans do not default to Haiku 4.5.
The photo editing app’s first “AI Lab” feature can automatically remove objects or people by matching the surrounding area, similar to removal tools in Photoshop and Google Photos.
An image upscaler is also “launching soon” that enhances resolution and details “while preserving color accuracy and composition,” according to VSCO.
As OpenAI’s Sam Altman reveals the company is ready to let ChatGPT sext with verified adult users, one commenter points out that this may create some ambiguity in the future.
Lewise:
Soon we’ll need some disambiguation when OpenAI reports how many subs it has
Get the day’s best comment and more in my free newsletter, The Verge Daily.
OpenAI wants to spend more than $1 trillion on its quest to build superintelligent AI, according to the Financial Times.
The figure dwarfs the roughly $13 billion it takes in each year, so the startup is counting on new shopping tools, consumer gadgets, online advertising, sales from Sora, and government deals to get it there. “Creative” debt arrangements are also on the table.
[Financial Times]
After rolling out Instant Checkout for Etsy purchases in September, OpenAI is expanding the feature to Walmart, allowing you to complete your purchase on the retail giant’s website from within ChatGPT.
Oracle will begin offering cloud services using AMD’s new Instinct M1450 Series GPUs next year, before expanding the rollout in 2027 and “beyond” as growing AI usage eats up more compute. The deal comes just days after OpenAI announced a deal to use AMD’s chips in its AI data centers.
The number of new online articles generated by AI hovers at around 50 percent, according to research from SEO firm Graphite. That figure, first reported by Axios, seems stable(ish) for now. Maybe there is hope that AI won’t crowd out real humans across the internet.
Correction: Replaced the graph with an error-free version.
Keep your eyes peeled for a new photo feature coming to Google Search soon. Google says Create mode will let you transform images with AI, courtesy of its popular image editor Nano Banana.
Nano Banana is also coming to NotebookLM’s Video Overviews and, eventually, Photos.
Tulloch helped launch Thinking Machines with Mira Murati, who famously left OpenAI to start her own company. Mark Zuckerberg’s attempts to buy Thinking Machines and other AI companies have been a pretty open secret. While the big swings to buy companies like Perplexity have fallen short, he has managed to steal away some high profile talent from rivals.
[The Wall Street Journal]
