AI Taught Robots to Build Computers — While Most Americans Wish It Would Slow Down
Today’s AI news tells two very different stories. Researchers are pushing the technology further than ever, with robots now learning to handle delicate hardware on their own. Meanwhile, a new poll shows most Americans aren’t celebrating — they’re worried. Both stories are worth your attention.
Nvidia’s New System Lets AI Teach Robots to Do Your Most Annoying Tech Tasks
Robots that can install a graphics card or snip a zip tie might sound like a niche party trick, but the system behind it is a significant shift in how robots learn. Nvidia, working with researchers at Carnegie Mellon and UC Berkeley, released an open-source framework called ENPIRE that uses AI coding agents — software programs that can write and revise their own code — to train robots on physical tasks without a human programming each step by hand. Ars Technica has the full breakdown.
Think of it like hiring a new employee who not only figures out the job on their own but also rewrites the training manual as they go. The AI agent observes what the robot is doing, identifies what’s going wrong, and adjusts the code driving the robot’s behavior. This loop runs continuously, with no engineer needing to step in between attempts.
For everyday people, the immediate impact is still a few years away. But this kind of self-improving system is what makes the leap from “robot that does one thing in a factory” to “robot that can handle a variety of household or workplace tasks” actually plausible. Installing a GPU requires precise hand movements and spatial awareness — skills that have historically required enormous amounts of manual programming to replicate.
Why this matters: If robots can teach themselves physical skills, the bottleneck of manually programming every task starts to disappear. That changes the timeline for useful home and workplace robots considerably.
“ENPIRE framework lets AI coding agents run the full loop of teaching robots practical tasks.”
Google Docs Keeps Pushing Its AI Tool on You — Here’s How to Make It Stop
If you’ve opened a Google Doc recently and been greeted by a pop-up urging you to try Gemini — Google’s AI writing assistant — you’re not alone, and you’re not imagining it. Google has been displaying these prompts to users, and enough people found them irritating that TechCrunch published a step-by-step guide on how to turn them off.
Dismissing the pop-up doesn’t make it go away permanently. To actually stop it, you need to dig into your Google Docs settings and disable the Gemini suggestions toggle. The option exists — it’s just not obvious, which is probably by design.
This matters beyond the minor annoyance. When a company embeds a feature so persistently into a free tool that millions rely on for work and school, it raises a reasonable question: whose experience is being prioritized? Google wants users to adopt Gemini. Users want to write a grocery list without being interrupted. For now, you can opt out — but knowing the setting exists requires knowing to look for it.
Why this matters: The ability to opt out of AI features is increasingly important as tech companies fold them into tools people use every day. This one’s fixable in two minutes.
Most Americans Are Using AI — and Most Are Uneasy About It
Here’s a number that should make the AI industry pause: just 16 percent of Americans believe AI will have a positive impact on society, according to a new Pew Research Center poll. TechCrunch covered the findings in detail. That’s a striking gap between public sentiment and the enthusiasm pouring out of Silicon Valley boardrooms.
The poll also found that roughly half of Americans now use AI chatbots — tools like ChatGPT or Gemini that you can hold a text conversation with — so this skepticism isn’t coming from people who’ve never touched the technology. About 63 percent of respondents said AI is advancing too fast. People aren’t unfamiliar with AI. They’re familiar with it and still concerned.
For anyone working in tech, selling products, or making policy decisions, this gap is a real problem. Trust is hard to earn and easy to lose. If the people AI is supposed to benefit don’t believe it will, adoption stays shallow and regulation becomes more likely to be reactive than thoughtful.
Why this matters: Public trust shapes how AI gets built, regulated, and used. A 16 percent confidence rating is a loud signal that the industry’s optimism isn’t landing.
“Only 16 percent of Americans think AI will have a positive impact on society.”
Also Happening in AI
Google is back in the smart speaker business after a six-year gap — its new Gemini-powered device is available now for $99.99, according to Wired. Separately, Snap released AR glasses called Specs for $2,195, featuring built-in cameras and computing; The Verge took a close look at whether anyone can actually pull them off. On the environmental front, Anthropic became the first AI startup to join the Frontier coalition, a group that funds carbon dioxide removal projects, per TechCrunch. And Roelof Botha — a former Sequoia Capital managing partner — joined SpaceX’s board of directors, adding another major tech figure to Elon Musk’s orbit.
What to Watch
The Pew numbers deserve more attention than a single news cycle. Watch whether AI companies respond with transparency efforts or simply push forward — how they react will reveal a lot about their priorities. On the robotics side, keep an eye on when ENPIRE-style self-training systems move out of research labs and into commercial products. That transition, whenever it happens, will be the real story.