Connect with us

Technologies

Bing AI Bungles Search Results at Times, Just Like Google

Expect search engines to tread carefully as tests reveal problems with AI language skills.

A close reading of Bing’s AI-boosted search results reveals that the website can make the same kinds of errors that are apparent in the ChatGPT technology foundation it uses and in Google’s competing Bard.

A new version of Bing that’s in limited testing employs large language model, or LLM, technology from OpenAI, the research lab Microsoft invested in and that grabbed the tech spotlight with its ChatGPT AI chatbot. ChatGPT and related technology, trained on enormous swaths of the internet, can produce remarkable results, but they don’t truly know facts and they can make errors.

Bing tries to avoid such errors by «grounding» results with Microsoft’s search technology, assessing the veracity and authority of source documents and offering links to sources, so people can evaluate the results on their own better. But AI and search engine researcher Dmitri Brereton and others have spotted errors in Bing’s demo, including flawed financial data from a Gap quarterly earnings report.

It’s a buzzkill moment for AI. The technology truly can produce remarkable and useful results, but the trouble is assessing when it isn’t doing that. Expect more caution as the search engine industry tries to find the right formula.

Google felt the pain after a demo last week when its Bard tool, which isn’t yet publicly available, produced erroneous information about the James Webb Space Telescope.

Microsoft offered a similar response Tuesday to what Google said about its gaffe: «We recognize that there is still work to be done and are expecting that the system may make mistakes during this preview period, which is why the feedback is critical so we can learn and help the models get better.»

One basic problem is that large language models, even if trained using text like academic papers and Wikipedia entries that have passed some degree of scrutiny, don’t necessarily assemble factual responses from that raw material.

As internet pioneer and Google researcher Vint Cerf said Monday, AI is «like a salad shooter,» scattering facts all over the kitchen but not truly knowing what it’s producing. «We are a long way away from the self-awareness we want,» he said in a talk at the TechSurge Summit.

Summarizing documents might appear to be within AI’s advanced language processing abilities, but constructing human-readable sentences without drawing in inappropriate information can be difficult. For example, in this request to summarize a spec sheet for Canon’s new R8 mirrorless camera, Bing technology rattled off many features that actually are found in Canon’s earlier R5.

Another high-profile computer scientist, former Stanford University Professor John Hennessy, was more bullish than Cerf, praising AI for its language skills in understanding search queries and generating results that often are correct. But he also cautioned, «It’s always confident that it has the right answer even when it doesn’t.»

OpenAI itself offers the same warning. «ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging,» the AI research lab said when it launched ChatGPT in November.

Editors’ note: CNET is using an AI engine to create some personal finance explainers that are edited and fact-checked by our editors. For more, see this post.

Technologies

How to Get Verizon’s New Internet Plan for Just $25 Per Month

Continue Reading

Technologies

This $20K Humanoid Robot Promises to Tidy Your Home. But There Are Strings Attached

The new Neo robot from 1X is designed to do chores. It’ll need help from you — and from folks behind the curtain.

It stands 5 feet, 6 inches tall, weighs about as much as a golden retriever and costs near the price of a brand-new budget car. 

This is Neo, the humanoid robot. It’s billed as a personal assistant you can talk to and eventually rely on to take care of everyday tasks, such as loading the dishwasher and folding laundry. 

Neo doesn’t work cheap. It’ll cost you $20,000. And even then, you’ll still have to train this new home bot, and possibly need a remote assist as well.

If that sounds enticing, preorders are now open (for a mere $200 down). You’ll be signing up as an early adopter for what Neo’s maker, a California-based company called 1X, is calling a «consumer-ready humanoid.» That’s opposed to other humanoids under development from the likes of Tesla and Figure, which are, for the moment at least, more focused on factory environments. 

Neo is a whole order of magnitude different from robot vacuums like those from Roomba, Eufy and Ecovacs, and embodies a long-running sci-fi fantasy of robot maids and butlers doing chores and picking up after us. If this is the future, read on for more of what’s in store.


Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.


What the Neo robot can do around the house

The pitch from 1X is that Neo can do all manner of household chores: fold laundry, run a vacuum, tidy shelves, bring in the groceries. It can open doors, climb stairs and even act as a home entertainment system.

Neo appears to move smoothly, with a soft, almost human-like gait, thanks to 1X’s tendon-driven motor system that gives it gentle motion and impressive strength. The company says it can lift up to 154 pounds and carry 55 pounds, but it is quieter than a refrigerator. It’s covered in soft materials and neutral colors, making it look less intimidating than metallic prototypes from other companies.

The company says Neo has a 4-hour runtime. Its hands are IP68-rated, meaning they’re submersible in water. It can connect via Wi-Fi, Bluetooth and 5G. For conversation, it has a built-in LLM, the same sort of AI technology that powers ChatGPT and Gemini.

The primary way to control the Neo robot will be by speaking to it, just as if it were a person in your home.  

Still, Neo’s usefulness today depends heavily on how you define useful. The Wall Street Journal’s Joanna Stern got an up-close look at Neo at 1X’s headquarters and found that, at least for now, it’s largely teleoperated, meaning a human often operates it remotely using a virtual-reality headset and controllers. 

«I didn’t see Neo do anything autonomously, although the company did share a video of Neo opening a door on its own,» Stern wrote last week. 

1X CEO Bernt Børnich told her that Neo will do most things autonomously in 2026, though he also acknowledged that the quality «may lag at first.»

The company’s FAQ says that for any chore request Neo doesn’t know how to accomplish, «you can schedule a 1X Expert to guide it» to help the robot «learn while getting the job done.»

What you need to know about Neo and privacy

Part of what early adopters are signing up for is to let Neo learn from their environment so that future versions can operate more independently. 

That learning process raises privacy and trust questions. The robot uses a mix of visual, audio and contextual intelligence — meaning it can see, hear and remember interactions with users throughout their homes. 

«If you buy this product, it is because you’re OK with that social contract,» Børnich told the Journal. «It’s less about Neo instantly doing your chores and more about you helping Neo learn to do them safely and effectively.»

Neo’s reliance on human operation behind the scenes prompted a response from John Carmack, a computer industry luminary known for his work with VR systems and the lead programmer of classic video games including Doom and Quake. 

«Companies selling the dream of autonomous household humanoid robots today would be better off embracing reality and selling ‘remote operated household help’,» he wrote in a post on the X social network (formerly Twitter) on Monday.

1X says it’s taking steps to protect your privacy: Neo listens only when it recognizes it’s being addressed, and its cameras will blur out humans. You can restrict Neo from entering or viewing specific areas of your home, and the robot will never be teleoperated without owner approval, the company says. 

But inviting an AI-equipped humanoid to observe your home life isn’t a small step.

The first units will ship to customers in the US in 2026. There is a $499 monthly subscription alternative to the $20,000 full-purchase price, though that will be available at an unspecified later date. A broader international rollout is promised for 2027.

Neo’s got a long road ahead of it to live up to the expectations set by Rosie the Robot in The Jetsons way back when. But this is no Hanna-Barbera cartoon. What we’re seeing now is a much more tangible harbinger of change.

Continue Reading

Technologies

I Wish Nintendo’s New Switch 2 Zelda Game Was an Actual Zelda Game

Hyrule Warriors: Age of Imprisonment has great graphics, a great story and Zelda is actually in it. But the gameplay makes me wish for another true Zelda title instead.

I’ve never been a Hyrule Warriors fan. Keep that in mind when I say that Nintendo’s new Switch 2-exclusive Zelda-universe game has impressed me in several ways, but the gameplay isn’t one of them. Still, this Zelda spinoff has succeeded in showing off the Switch 2’s graphics power. Now can we have a true Switch 2 exclusive Zelda game next?

The upgraded graphics in Tears of the Kingdom and Breath of the Wild has made the Switch 2 a great way to play recent Zelda games, which had stretched the Switch’s capabilities to the limit before. And they’re both well worth revisiting, because they’re engrossing, enchanting, weird, epic wonders. Hyrule Warriors: Age of Imprisonment, another in the Koei-Tecmo developed spinoff series of Zelda-themed games, is a prequel to Tears of the Kingdom. It’s the story of Zelda traveling back in time to ancient Hyrule, and the origins of Ganondorf’s evil. I’m here for that, but a lot of hack and slash battles are in my way. 

A handful of hours in, I can say that the production values are wonderful. The voices and characters and worlds feel authentically Zelda. I feel like I’m getting a new chapter in the story I’d already been following. The Switch 2’s graphics show off smooth animation, too, even when battles can span hundreds of enemies.

But the game’s central style, which is endless slashing fights through hordes of enemies, gets boring for me. That’s what Hyrule Warriors is about, but the game so far feels more repetitive than strategic. And I just keep button-mashing to get to the next story chapter. For anyone who’s played Hyrule Warriors: Age of Calamity, expect more of the same, for the most part.

I do like that the big map includes parts in the depths and in the sky, mirroring the tri-level appeal of Tears of the Kingdom. But Age of Calamity isn’t a free-wandering game. Missions open up around the map, each one opening a contained map to battle through. Along the way, you unlock an impressive roster of Hyrule characters you can control.

As a Switch 2 exclusive to tempt Nintendo fans to make the console upgrade, it feels like a half success. I admire the production values, and I want to keep playing just to see where the story goes. But as a purchase, it’s a distant third to Donkey Kong Bananza and Mario Kart World.

Hyrule Warriors fans, you probably know what you’re probably in for, and will likely get this game regardless. Serious Zelda fans, you may enjoy it just for the story elements alone. 

As for me? I think I’ll play some more, but I’m already sort of tuning the game out a bit. I want more exploration, more puzzles, more curiosity. This game’s not about that. But it does show me how good a true next-gen Zelda could be on the Switch 2, whenever Nintendo decides to make that happen.

Continue Reading

Trending

Copyright © Verum World Media