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

Google I/O 2025: How to Watch and What to Expect

With Android 16 out of the way, Google I/O will certainly be all about AI.

Google I/O 2025 takes place on May 20 and 21 with Google’s big keynote happening on day 1. We expect Big G to talk about its myriad innovations across its ever-expanding portfolio of products — almost certainly with a huge focus on AI every step of the way. If we collectively cross our fingers, promise to be good and eat all our vegetables then we may even be treated to a sneak peek at upcoming hardware.

Read more: Android 16: Everything Google Announced at the Android Show

Google also hosted a totally separate event that focused solely on Android. The Android Show: I/O Edition saw the wrappers come off Android 16, with insights into the new Material 3 Expressive interface, updates to security and a focus on Gemini and how it’ll work on a variety of other devices.

By breaking out Android news into its own virtual event, Google frees itself to spend more time during the I/O keynote to talk about Gemini, Deep Mind, Android XR and Project Astra. It’s going to be a jam-packed event, so here’s how you can watch I/O 2025 as it happens and what you can look forward to.

Google I/O: Where to watch

Google I/O proper kicks off with a keynote taking place on May 20, 10 a.m. PDT (1 p.m. EDT, 6 p.m. BST). It’ll almost certainly be available to stream online on Google’s own YouTube channel, although a holding video is yet to be available. There’s no live link on the I/O website yet, either, though you can use the handy links to add the event to your calendar of choice. Expect links to a livestream to be available closer to the day.

What to expect from Google I/O 2025

Little chat about Android 16: As Google gave Android 16 its own outing already, it’s likely that it won’t be mentioned all that much during I/O. In fact at last year’s event, Android was barely mentioned, while uses of the term «AI» went well over a hundred.

Android XR: Google didn’t talk much about Android XR during the Android show, focusing instead on the purely phone-based updates to the platform. We expected to hear more about the company’s latest foray into mixed-reality headsets in partnership with Samsung and its Project Moohan headset, so it’s possible that this is being saved for I/O proper.

Gemini: With Android being spun out into its own separate event, Google is evidently clearing the way for I/O to focus on everything else the company does. AI will continue to dominate the conversation at I/O, just as it did last year (though hopefully Google can make it more understandable) with updates to many of its AI platforms expected to be announced.

Gemini is expected to receive a variety of update announcements, including more information on its latest 2.5 Pro update which boasts various improvements to its reasoning abilities, and in particular to its helpfulness for coding applications. Expect lots of mentions of Google’s other AI-based products, too, including DeepMind, LearnLM and Project Astra. Let’s just hope Google has figured out how to make this information make any kind of sense.

Beyond AI, Google may talk about updates to its other products including GMail, Chrome and the Play Store, although whether these updates are big enough to be discussed during the keynote rather than as part of the developer-focused sessions following I/O’s opening remains to be seen.

Technologies

You Can Now Buy Nike’s $900 Workout Shoes for Compression and Heating

The Nike Hyperboots, designed to help you warm up and recover from workouts, launched Saturday.

Those workout shoes with compression and heating that Nike and Hyperice showed off at CES 2025 earlier this year weren’t just a concept. The Hyperboot is now available to buy online in North America, so they’re within reach, as long as you’re willing to spend $899.

The high-tops, which Nike and Hyperice call a wearable much like your smartwatch, help your feet warm up before a workout, and then recover after it. The shoes do this with heating and air-compression massage technology, taking the idea of heating pads and compression socks and making them mobile.

«You can definitely feel the heat in here,» CNET former mobile senior writer Lisa Eadicicco said when she had the chance to try these workout shoes on in January. She walked across a demo room in Las Vegas wearing the fancy footwear to test out the compression and heating features.

The boots massage and compress your ankles and feet, and in CNET’s test, we could especially feel the heat around the ankles. Buttons on the shoes let you adjust compression and the amount of heat with multiple settings for each.

«The Hyperboot contains a system of dual-air bladders that deliver sequential compression patterns and are bonded to thermally efficient heating elements that evenly distribute heat throughout the shoe’s entire upper,» Nike said.

The battery lasts for 1 to 1.5 hours on max heat and compression settings, or 8 hours if you’re only using the massage setting. It takes 5 to 6 hours to charge via USB-C cable. The boots come in five sizes: S, M, L, XL and XXL.