An attendee interacts with the AI-powered Microsoft Bing search engine and Edge browser during an event at the company’s headquarters in Redmond, Washington, US, on Tuesday, Feb. 7, 2023. Microsoft unveiled new versions of its Bing internet-search engine and Edge browser powered by the newest technology from ChatGPT maker OpenAI.
Chona Kasinger | Bloomberg | Getty Images
Microsoft has given a small group of people early access to the new version of its Bing search engine boosted with artificial intelligence courtesy of startup OpenAI, the maker of ChatGPT.
CNBC has spent some time testing it. The new Bing can at times be more helpful, or at least more entertaining, than the usual blue links in search results. And it’s similar to ChatGPT in that it provides a lot more information than you might expect from a traditional search.
If Microsoft manages to get more people to use Bing, it could make the company even more profitable than it already is. For every percentage point that Microsoft gains in search advertising, it will pick up $2 billion in new revenue, Phil Ockenden, finance chief for the company’s Windows, devices and search divisions, said on a Tuesday conference call with analysts. “This is the largest software category that exists, and it’s incredibly profitable, incredibly large and still growing,” Amy Hood, Microsoft’s chief financial officer, said on the call.
So far, the new Bing feels like it’s been supercharged, and at the very least, people might want to try it out to see if it satisfies them more than traditional search engines that billions of people have come to know in the past 25 years.
Here’s what it’s like.
You can chat with the new Bing
After you search on Bing, you can challenge the results rather than clicking on a few URLs or typing out a new query. To compare, I asked the current version of Bing to identify the largest software category, to which it said the answer is “enterprise software” with a citation to Statista. The new version provides similar information at the top of the search results page, but below that, you’ll find a text box in which you can type a message and kick off a chat. You might ask, “Really?” And Bing will respond with more information attempting to validate its previous answer.
That gets into the question of accuracy. You might ask the AI-boosted search engine if the response is wrong, for example. And the new chat feature will say that “one could argue that search advertising is the largest software category in the world by revenue,” and hedges by noting there are many ways to evaluate different kinds of software. That’s not what we’re used to seeing when we go to a search engine. It’s downright entertaining.
Bing is now like ChatGPT but it can provide more information
The chat feature in Bing can also perform a variety of fun maneuvers that people have come to know they can do with ChatGPT, the OpenAI chatbot that’s been available since late November. And people will surely compare it with Bing’s new chatbot.
With some queries, Bing, drawing on OpenAI’s GPT AI model, provides results that appear to go above and beyond what was asked compared to ChatGPT.
Consider the following prompt: “If I wanted to familiarize myself with the concept of German expressionism, what movies, music, and literature should I watch, listen to, and read?”‘
When that prompt is entered into ChatGPT, the OpenAI tool generates three bulleted lists detailing examples of German Expressionism in film, music and literature. The bullet points are simple and economical, containing examples of German Expressionism such as the 1920 film “The Cabinet of Dr. Caligari” and Gustav Meyrink’s novel “The Golem.”
Bing not only presents lists of cinema, music, and literature representing German Expressionism, but it also gives users extra context about the artistic movement. The result looks like a Wikipedia entry about German Expressionism, complete with footnotes linking to the source material, coupled with examples of the genre that complied with the prompt’s request.
Other differences between Bing and ChatGPT
Microsoft’s supercharged Bing seems to offer better advice than ChatGPT, at least regarding the following prompt: “Create a fitness routine and meal plan for me over the next 3 months. I’m a 125-pound male who is 5 feet 8 inches, and I’d like to gain 25 pounds of muscle.”
When given that prompt, ChatGPT displays a bulleted list of a proposed fitness routine and meal plan that presumably would lead someone to gain 25 pounds of muscle in 90 days. The tips include weightlifting (45-60 minutes, four or five times a week), cardio (20-30 minutes, two to three times a week) and a dinner “that is high in protein, healthy fats, and complex carbohydrates. Examples include salmon with quinoa and vegetables or a turkey burger with sweet potato fries.”
Bing, however, notes that it might not be realistic to gain 25 pounds in three months, and it warns that doing so could be “potentially unhealthy.” Gaining that much muscle mass could “require a lot of genetic potential, steroids, or both,” Bing noted, linking out to an article from the Healthline website about the subject.
Recognizing that the search query contains a potentially harmful premise, Bing suggests that you “adjust your expectations and aim for a more reasonable and sustainable goal, such as 10-15 pounds of muscle in 3 months.”
The search tool then shares a list of some general tips to help people bulk up, including adding more protein to one’s diet, lifting weights and getting rest.
At times the Bing chatbot opted not to do things that ChatGPT would do. Bing demurred when asked to compose an email to employees telling them that some would be laid off, while others would be given recognition for their excellent performance. But OpenAI gladly created an email, subject line and all.
Bing said that coming up with such a message is “a sensitive and personal matter that requires human judgment and empathy.”
Both Microsoft and OpenAI have made it clear that their chatbots can sometimes provide inaccurate information, and CNBC encountered that on both. Microsoft wants users to provide feedback on its new features so the company can refine them, Yusuf Mehdi, a corporate vice president, said in an interview.
And when you don’t receive the sort of response you have in mind after chatting with Bing for a while, a button with a broom icon near the bottom of the window might be able to help. Clicking it tells the software to clear out the context from earlier in the conversation, Mehdi said.
Microsoft hasn’t opened up the enhanced Bing to everyone yet. People can sign up by joining a waiting list, although it currently requires people to install the Edge browser for MacOS or Windows.
Sanjay Beri, chief executive officer and founder of Netskope Inc., listens during a Bloomberg West television interview in San Francisco, California.
David Paul Morris | Bloomberg | Getty Images
Cloud security platform Netskope will go public on the Nasdaq under the ticker symbol “NTSK,” the company said in an initial public offering filing Friday.
The Santa Clara, California-based company said annual recurring revenue grew 33% to $707 million, while revenues jumped 31% to about $328 million in the first half of the year.
But Netskope isn’t profitable yet. The company recorded a $170 million net loss during the first half of the year. That narrowed from a $207 million loss a year ago.
Netskope joins an increasing number of technology companies adding momentum to the surge in IPO activity after high inflation and interest rates effectively killed the market.
So far this year, design software firm Figma more than tripled in its New York Stock Exchange debut, while crypto firm Circle soared 168% in its first trading day. CoreWeave has also popped since its IPO, while trading app eToro surged 29% in its May debut.
Read more CNBC tech news
Netskope’s offering also coincides with a busy period for cybersecurity deals.
Founded in 2012, Netskope made a name for itself in its early years in the cloud access security broker space. The company lists Palo Alto Networks, Cisco, Zscaler, Broadcom and Fortinet as its major competitors.
Netskope’s biggest backers include Accel, Lightspeed Ventures and Iconiq, which recently benefited from Figma’s stellar debut.
Morgan Stanley and JPMorgan are leading the offering. Netskope listed 13 other Wall Street banks as underwriters.
Meta CEO Mark Zuckerberg makes a keynote speech at the Meta Connect annual event at the company’s headquarters in Menlo Park, Calif., on Sept. 25, 2024.
Manuel Orbegozo | Reuters
Meta is planning to use its annual Connect conference next month to announce a deeper push into smart glasses, including the launch of the company’s first consumer-ready glasses with a display, CNBC has learned.
That’s one of the two new devices Meta is planning to unveil at the event, according to people familiar with the matter. The company will also launch its first wristband that will allow users to control the glasses with hand gestures, the people said.
Connect is a two-day conference for developers focused on virtual reality, AR and the metaverse. It was originally called Oculus Connect and obtained its current moniker after Facebook changed its parent company name to Meta in 2021.
The glasses are internally codenamed Hypernova and will include a small digital display in the right lens of the device, said the people, who asked not to be named because the details are confidential.
The device is expected to cost about $800 and will be sold in partnership with EssilorLuxottica, the people said. CNBC reported in October that Meta was working with Luxottica on consumer glasses with a display.
Meta declined to comment. Luxottica, which is based in France and Italy, didn’t respond to a request for comment.
Meta began selling smart glasses with Luxottica in 2021 when the two companies released the first-generation Ray-Ban Stories, which allowed users to take photos or videos using simple voice commands. The partnership has since expanded, and last year included the addition of advanced AI features that made the second generation of the product an unexpected hit with early adopters.
Luxottica owns a number of glasses brands, including Ray-Ban, and licenses many others like Prada. It’s unclear what brand Luxottica will use for the glasses with AR, but a Meta job listing posted this week said the company is looking for a technical program manager for its “Wearables organization,” which “is responsible for the Ray-Ban AR glasses and other wearable hardware.”
In June, CNBC reported that Meta and Luxottica plan to release Prada-branded smart glasses. Prada glasses are known for having thick frames and arms, which could make them a suitable option for the Hypernova device, one of the people said.
Last year, Meta CEO Mark Zuckerberg used Connect to showcase the company’s experimental Orion AR glasses.
The Orion features AR capabilities on both lenses, capable of blending 3D digital visuals into the physical world, but the device served only as a prototype to show the public what could be possible with AR glasses. Still, Orion built some positive momentum for Meta, which since late 2020 has endured nearly $70 billion in losses from its Reality Labs unit that’s in charge of building hardware devices.
With Hypernova, Meta will finally be offering glasses with a display to consumers, but the company is setting low expectations for sales, some of the sources said. That’s because the device requires more components than its voice-only predecessors, and will be slightly heavier and thicker, the people said.
Meta and Ray-Ban have sold 2 million pairs of their second-generation glasses since 2023, Luxottica CEO Francesco Milleri said in February. In July, Luxottica said that revenue from sales of the smart glasses had more than tripled year over year.
As part of an extension agreement between Meta and Luxottica announced in September, Meta obtained a stake of about 3% in the glasses company according to Bloomberg. Meta also gets exclusive rights to Luxottica’s brands for its smart glasses technology for a number of years, a person familiar with the matter told CNBC in June.
Although Hypernova will feature a display, those visual features are expected to be limited, people familiar with the matter said. They said the color display will offer about a 20 degree field of view — meaning it will appear in a small window in a fixed position — and will be used primarily to relay simple bits of information, such as incoming text messages.
Andrew Bosworth, Meta’s technology chief, said earlier this month that there are advantages to having just one display rather than two, including a lower price.
“Monocular displays have a lot going for them,” Bosworth said in an Instagram video. “They’re affordable, they’re lighter, and you don’t have disparity correction, so they’re structurally quite a bit easier.”
‘Interact with an AI assistant’
Other details of Meta’s forthcoming glasses were disclosed in a July letter from the U.S. Customs and Border Patrol to a lawyer representing Meta. While the letter redacted the name of the company and the product, a person with knowledge of the matter confirmed that it was in reference to Meta’s Hypernova glasses.
“This model will enable the user to take and share photos and videos, make phone calls and video calls, send and receive messages, listen to audio playback and interact with an AI assistant in different forms and methods, including voice, display, and manual interactions,” according to the letter, dated July 23.
The letter from CBP was part of routine communication between companies and the U.S. government when determining the country of origin for a consumer product. It refers to the product as “New Smart Glasses,” and says the device will feature “a lens display function that allows the user to interface with visual content arising from the Smart Features, and components providing image data retrieval, processing, and rendering capabilities.”
CBP didn’t provide a comment for this story.
The Hypernova glasses will also come paired with a wristband that will use technology built by Meta’s CTRL Labs, said people familiar with the matter. CTRL Labs, which Meta acquired in 2019, specializes in building neural technology that could allow users to control computing devices using gestures in their arms.
The wristband is expected to be a key input component for the company’s future release of full AR glasses, so getting data now with Hypernova could improve future versions of the wristband, the people said. Instead of using camerasensors to track body movements, as with Apple’s Vision Pro headset, Meta’s wristband uses so-called sEMG sensortechnology, which reads and interprets the electrical signals from hand movements.
One of the challenges Meta has faced with the wristband involves how people choose to wear it, a person familiar with the product’s development said. If the device is too loose, it won’t be able to read the user’s electrical signals as intended, which could impact its performance, the person said. Also, the wristband has run into issues in testing related to which arm it’s worn on, how it works on men versus women and how it functions on people who wear long sleeves.
The CTRL Labs team published a paper in Nature in July about its wristband, and Meta wrote about it in a blog post. In the paper, the Meta team detailed its use of machine learning technology to make the wristband work with as many people as possible. The additional data collected by the upcoming device should improve those capabilities for future Meta smart glasses.
“We successfully prototyped an sEMG wristband with Orion, our first pair of true augmented reality (AR) glasses, but that was just the beginning,” Meta wrote in the post. “Our teams have developed advanced machine learning models that are able to transform neural signals controlling muscles at the wrist into commands that drive people’s interactions with the glasses, eliminating the need for traditional—and more cumbersome—forms of input.”
Bloomberg reported the wristband component in January.
Meta has recently started reaching out to developers to begin testing both Hypernova and the accompanying wristband, people familiar with the matter said. The company wants to court third-party developers, particularly those who specialize in generative AI, to build experimental apps that Meta can showcase to drum up excitement for the smart glasses, the people said.
In addition to Hypernova and the wristband, Meta will also announce a third-generation of its voice-only smart glasses with Luxottica at Connect, one person said.
That device was also referenced by CBP in its July letter, referring to it as “The Next Generation Smart Glasses.” The glasses will include “components that provide capacitive touch functionality, allowing users to interact with the Smart Glasses through touch gestures,” the letter said.
Google CEO Sundar Pichai gestures to the crowd during Google’s annual I/O developers conference in Mountain View, California on May 20, 2025.
Camille Cohen | Afp | Getty Images
Alphabet shares rose on a Friday report that Apple is in early discussions to use Google’s Gemini AI models for an updated version of the iPhone-maker’s Siri assistant.
The company’s shares rose more than 3% on the Bloomberg report, which said Apple recently inquired of Google about the potential for the search giant to build a custom AI model that would power a new Siri that could launch next year. Google’s flagship AI models Gemini have consistently been atop key benchmarks for artificial intelligence advancements while Apple has struggled to define its own AI strategy.
The reported talks come as Google faces potential risk to its lucrative search deals with Apple. This month, a U.S. judge is expected to rule on the penalties for Google’s alleged search monopoly, in which the Department of Justice recommending eliminating exclusionary agreements with third parties. For Google, that refers to its search position on Apple’s iPhone and Samsung devices — deals that cost the company billions of dollars a year in payouts.
The Android maker has said its Gemini models will become the default assistant on Android phones. Google this year has showed Gemini doing capabilities that go beyond Siri’s capabilities, such as summarizing videos.
Craig Federighi, who oversees Apple’s operating systems, said at last year’s developer conference that the iPhone maker would like to add other AI models for specific purposes into its Apple Intelligence framework. Federighi specifically mentioned Google, whose Gemini can now hold conversations with users and handle input that comes from photos, videos, voice or text. Apple is also exploring partnerships with Anthropic and OpenAI as it tried to renew its AI roadmap, according to a June Bloomberg report.
Documents revealed during Google’s remedy trial showed executives from Apple were involved in the negotiations over using Google’s Gemini for a potential search option.