Bumble is testing an AI dating assistant called "Bee" that it hopes will get users on dates without them having to swipe through profiles, Bloomberg writes. The company announced the AI assistant during its fourth quarter earnings, and intends to use the AI in a new experience it calls "Dates."When a user opts in to Bumble's Dates feature, Bee performs an onboarding chat where it learns about the users' "values, relationship goals, communications style, lifestyle and dating intentions," and then attempts to find other users who share some or all of those traits. Once Bee finds someone compatible, both users are notified in the app that they could be a great match, and receive a summary generated by Bee explaining why. From there, they can chat and see if things lead to a real-life date.As is often the case with pie-in-the-sky AI features, Bumble has even bigger plans for how Bee could be used in its app, including as a tool for collecting anonymous feedback from user's previous matches or as a way to receive suggestions for dates ideas. AI will also apparently enable Bumble to move away from binary yes or no swipes on profiles and towards a system where users connect over "chapter-based" profiles that are more reflective of their life story.Bumble is testing Bee internally and plans to launch the AI and its Dates feature in beta soon. The company is far from the only dating app experimenting with integrating AI recommendations and summaries. Tinder uses AI to recommend profile pictures to users, and now offers another feature called "Chemistry" that combines insights gained from personal questions and access to users' Camera Roll to make more informed matches. Meanwhile, Grindr's "Edge" subscription tier offers AI summaries of past chats and connections, and stats on whether a user is actually compatible with a new match. It's too early to tell whether AI makes a meaningful difference in the dating experience for users, but if it keeps them using an app or paying for a subscription, it's likely a worthwhile experiment for Bumble, Tinder and Grindr.This article originally appeared on Engadget at https://www.engadget.com/apps/bumble-is-the-latest-dating-app-to-add-an-ai-assistant-181729994.html?src=rss
Ankers Soundcore projectors have become an attractive option for buyers thanks to models like the P1 and Nebula X1 that combine performance and portability. Now, the company has added "affordability" to that equation with its latest model, the $369 P1i. Instead of being detachable like on the P1, its speakers fold out toward listeners, promising better and louder sound than most cheap projectors.
The P1i also delivers 1080p video, Google TV for streaming and the same easy screen fit setup as other Anker projectors. However, unlike some portable models, it lacks a built-in battery for true portability. Overall brightness is lower as well at 380 lumens compared to 650 on the P1. Despite those issues, the P1i is an outstanding budget projector thats ideal for impromptu indoor screenings and outdoor party use.
Features and design
With its relatively small size (8.9 x 7.2 x 8.0 inches) and slight five pound weight, the Soundcore Nebula P1i can be perched nearly anywhere and is easily toted around by the soft handle on top. Its almost too light, as even a slight bump can displace it and ruin the picture. It does have a tripod mount, though, which makes it easy to set up outside or install on your ceiling.
Fortunately, the P1i is perhaps the easiest projector in this price range to align with your wall or screen. It has Ankers Smart Instant Setup that automatically adjusts the autofocus, keystone correction, screen fit and obstacle avoidance. For my 120-inch roll down screen, I just had to set the P1i about 11 feet from the wall, make sure it was fairly level and then tilt the stand so it pointed at the screen. Then, I just ran the smart setup which instantly gave me a correctly set-up image.
The P1i is not a laser projector but its LED light source lasts even longer, rated for over 30,000 hours. It runs fairly cool compared to a bulb or laser model, but it still has a fan that kicks on from time to time. One annoyance is that the fan seemed to engage even when the projector was turned off, so I found I had to unplug it to stop that.
Steve Dent for Engadget
Connectivity-wise, the P1i has a single HDMI 2.0 port with ARC (Audio Return Channel), along with a USB-A port for external media and a 3.5mm auxiliary audio output. The projector allows for a maximum input of 4K at 60 fps (it outputs 1080p at 60 fps), so its only good for light gaming. Plus, youll want to keep in mind that if you do any keystone correction, gaming latency quadruples from 20ms to 80ms. It also supports Wi-Fi 5 for streaming and Bluetooth 5.1 to connect extra speakers.
Google TV is built in for streaming and projector control, which is a nice feature for a projector under $400. It provides a large library of apps via Google Play along with a familiar interface. You also get Netflixs official application with support for 4K Dolby Vision without the need to plug in a streaming device, plus theres Chromecast support and Google Assistant for voice control. The downside is that the interface can occasionally be sluggish. As for those foldable speakers, you can aim them up, down, forward or backwards, and they support Dolby Audio at 10 watts each.
Image quality
Image quality is decent but not outstanding, about what youd expect for a projector at this price. While its very sharp in the center, the edges are fuzzier, particularly if youre tilting the projector up or down and using plenty of keystone correction. With just 380 lumens of brightness (I measured it at just over 300 lumens in standard mode), its best to use the P1i in a dark room or at night if theres any ambient light, you wont be able to see much detail.
The P1i offers multiple picture profiles, including Standard, Vivid, Sports, Movies and Games, along with a Conference mode for maximum brightness. I found that the Vivid setting gave me the best mix of color accuracy and brightness.
Steve Dent for Engadget
Color accuracy is a strong point, with natural looking, nicely saturated hues. Surprisingly, the P1i supports 10-bit HDR to help improve color accuracy and max out what little brightness there is, though its a gigantic stretch calling a 380 lumen image (around 50 nits) "HDR." I tested the P1i on my usual roster of content, including Once Upon a Time in Hollywood, Andor, Dune 2 and Spider-Man: No Way Home, and found the colors to be pleasing and realistic. If theyre not quite to your liking, you can make fine adjustments manually.
Sound
The primary selling point of the Nebula P1I is those fold out speakers. However, considering Soundcores reputation for solid audio performance (take the Nebula X1 Pro for example), sound quality was a bit disappointing. Its relatively tinny with very little bass, providing a less-than-theatrical experience. Those speakers can go pretty loud, though, and midrange sounds like voices are very clear.
Fortunately, there are a couple of solutions. The P1i has multiple EQ settings, including Movies (with enhanced bass and treble), Music for the most natural sound and Outdoors to ensure clarity in open and noisy spaces. The best of those, I found, was Movies, as it mildly improves the lows and highs. Theres also a custom mode that lets you fine-tune the equalization. I was able to get acceptable sound quality when I boosted the bass to nearly the maximum level.
The other option, of course, is to pair the P1i to some Bluetooth speakers. I did just that with Soundcores external speakers sold as an option with the Nebula X1 and the sound quality was much improved. Of course, that would remove much of its convenience and increase the price for better sound.
Wrap-up
Steve Dent for Engadget
Ankers $400 Soundcore Nebula P1i is an excellent option for buyers looking for a portable projector to use outdoors or in the place of a second TV, though it suffers from subpar sound and picture quality. Competitors like XGIMIs MoGo 2 Pro offer similar brightness, built-in speakers and Google TV. And if you have a bit more in your budget, Ankers $800 Nebula P1 is brighter and easier to set up. Finally, for those looking for battery portability for camping or other activities, take a look at XGIMIs $500 Halo+ or the $530 Anker Nebula Capsule.
The P1i is great for simple nighttime movie screenings outdoors. If you feel like watching a movie or series in bed, you can bring it into your room, project the picture onto a wall and enjoy your streaming service of choice. You may not be disappointed with the overall picture quality, though you may wish for slightly better sound.This article originally appeared on Engadget at https://www.engadget.com/home/home-theater/soundcore-nebula-p1i-projector-review-an-affordable-option-with-accurate-color-and-loud-sound-180034925.html?src=rss
Its already been a busy year for high-profile Microsoft departures, with longtime Xbox chief Phil Spencer bowing out last month alongside his expected successor Sarah Bond. Today its the turn of Microsoft's head of Experiences + Devices, Rajesh Jha, who leaves after more than 35 years at the company.
Jha, who oversaw some of Microsofts most important products and services, including Windows, Office and Teams, said in a press release that hes been planning for his succession alongside CEO Satya Nadella for a while. Rather than bringing in a direct replacement, four members of his team will be promoted to executive vice president and report directly to Nadella.
Taking up these new more senior roles are Office EVP and LinkedIn CEO Ryan Roslanksky, Windows and Surface president Pavan Davuluri, president of business and industry Copilot (BIC) Charles Lamanna, and Microsoft 365 Core chief Perry Clarke. The outgoing Jha, who said his long career at Microsoft had been "an incredible privilege," will officially transition out of Microsoft on July 1, after which hell remain in an advisory role.
"When I think about the pantheon of leaders who have truly shaped this company, Rajesh stands firmly among them," Nadella said in a statement, adding that Jha had been a "constant" during his own time at the company. "He embodies the commitment that helped build and transform Microsoft into the company it is today, and it is on the strength of that foundation that we will continue to move forward."This article originally appeared on Engadget at https://www.engadget.com/computing/another-longtime-microsoft-executive-is-retiring-174453788.html?src=rss
Amazon just unveiled a new personality type for Alexa+. The "sassy" option is reserved for adults and the company claims it will throw out censored curse words from time to time. Amazon describes this option as a combination of "unfiltered personality" and "razor-sharp wit, playful sarcasm and occasional censored profanity."
We aren't yet sure how the chatbot handles the censoring. Does it use a garden variety bleep or a replacement word like fudge or something? I managed to get it to say "damn" and "hell", but couldn't force anything more profane than that.
In any event, adult users have to jump through a couple of hoops to activate this mode. It won't work if there's an enabled Amazon Kids profile on the account and it requires additional security checks, like face scans. The company also warns people upon being selected that the new tone could contain "mature subject matter." I'm more afraid of the bot using "clever comebacks" to absolutely shred my buying habits. Yes, I buy bagged popcorn when I have plenty of uncooked kernels in the pantry. I'm working on it.
This is still Alexa+, despite the ability to drop colorful language every now and again. It's not an adult AI companion like the anime-inspired weirdness Grok recently trotted out or whatever erotica-infused nonsense OpenAI has been working on. Also, Amazon says the bot won't get involved with hate speech, illegal activities, personal attacks or anything that could cause harm.
Amazon
This is just the latest personality type that the company has introduced for the chatbot. Users can also choose from sweet, brief or chill, with the last one resembling a surfer archetype. Alexa+ is an updated version of the company's long-standing chatbot that prioritizes natural-sounding conversation. It's fine, more or less, but I still use it primarily for alarms and weather.This article originally appeared on Engadget at https://www.engadget.com/ai/alexa-can-now-swear-thanks-to-a-new-personality-style-172106310.html?src=rss
Fancy yourself as one of those folks who stands in front of an expensive touchscreen display on a news network on election night, zooming in and out of counties while bleating about polling and voting data? If so, you might get a kick out of BallotGuessr. This is a riff on GeoGuessr that tasks you with guessing how a county voted in the 2024 presidential election. All you have to go on to figure out the identity of each county are contextual clues from Google Street View images. You can move around the environment a bit, but unless you get lucky, you'll need to have a good sense of politics and geography to do well here.Once you think you have an idea of where the county is, you move a slider to guess whether residents voted for the Democrat or Republican ticket and by how many points. In the daily challenge mode, you only have 30 seconds to make your guess in each of five rounds. I'm bad at it, but it's a fun take on GeoGuessr all the same.BallotGuessr features 2,845 curated Google Street View locations from all 50 states, with a maximum of 15 locations for each county. Its creator plans to expand the game with data for the 2022 midterms and 2020 presidential election, as well as recent elections in France, Germany and the United Kingdom.This article originally appeared on Engadget at https://www.engadget.com/gaming/ballotguessr-is-geoguessr-for-budding-political-pundits-170028894.html?src=rss
Ukraine's four-year war with Russia has made it the world leader in battlefield drone technology. One byproduct of that is that the data it collects has become one of the country's most valuable assets. On Thursday, Ukraine played that card, saying it will begin sharing its battlefield data with allies to train drone AI software."In modern warfare, we must defeat Russia in every technological cycle," Ukraine Defense Minister Mykhailo Fedorov wrote on Telegram (translated from Ukrainian). "Artificial intelligence is one of the key areas of this competition."Fedorov previewed the move when he took his post in January. At the time, the tech-savvy cabinet member pledged to "more actively" bring allies into projects. Foreign allies and companies have sought access to the country's data as, for better or worse, AI increasingly becomes an integral element of warfare.Fedorov says Ukraine has a platform that will safely train partners' AI models without providing sensitive data. The system is said to provide continually updating datasets, including large volumes of photos and videos."For us, this is the next step in the development of win-win cooperation," Fedorov wrote. "Partners get the opportunity to train their AI models on real data from modern warfare. And [for] Ukraine: faster development of autonomous systems and new technological solutions for the front."Last year, Ukrainian President Volodymyr Zelenskyy warned global leaders of a dangerous escalation tied to drone tech and AI. We are now living through the most destructive arms race in human history, he said at a meeting of the UN General Assembly in September. However, given the ugly realities in his country, Zelenskyy reiterated his need for armaments. The only guarantee of security is friends and weapons, he said.This article originally appeared on Engadget at https://www.engadget.com/ai/ukraine-allows-allies-to-train-ai-models-on-its-battlefield-data-165104853.html?src=rss
After spending many years as the public face of Overwatch, Jeff Kaplan stayed well out of the limelight after leaving Blizzard in 2021. Five years later, the former Blizzard vice president and Overwatch lead director is back with his own studio and a new game, which you might be able to play pretty soon.The Legend of California is billed as an open-world, action-survival shooter. It looks like a mix of Red Dead Redemption and Rust (Rust Dead Redemption, if you will). It's set during the gold rush era, but Kaplan says he and his team at Kintsugiyama were not aiming for historical accuracy. For one thing, this version of California is an island. Still, the developers wanted to make the game feel authentic to the time period.There are cowboys and prospectors, and you'll be able to go hunting, build mines and stables, craft tools and weapons, build out your homestead and raid hostile camps. There are "challenging" player vs. environment encounters (Kaplan says there are four difficulty tiers available to start with) and optional player vs. player battles. You'll be able to form a company with up to three other players and share progress, resources, buildings and other things with them.Kaplan says his 34-strong team hand-crafted the world, though there's a degree of randomization at play. A certain biome (say, the game's version of the Mojave Desert) might be the easiest, most beginner-friendly area of the game on one server, and the endgame, tier four section on another. The points of interest might pop up in unexpected spots too an Alcatraz-inspired structure will appear in a Bay Area-style region in some world seeds, and in snowy mountains in others.The Legend of California is being published by Blizzard co-founder and former CEO Mike Morhaime's company Dreamhaven. It's slated to enter early access on Steam and the Epic Games Store later this year, with closed alpha playtests expected to start soon.Kaplan announced The Legend of California in unusual fashion. Not during a splashy showcase, but in a five-hour appearance on Lex Fridman's podcast. Kaplan discussed his life and career, including his work on World of Warcraft, the ill-fated Titan and, of course, Overwatch. In his first public appearance since stepping down as Overwatch director, Kaplan revealed his reasons for leaving Blizzard, where he spent 19 years and previously had no intention of leaving. Business pressures related to the Overwatch League and Overwatch 2 played a part in his departure, but the final straw came in 2020 during a meeting with Blizzard's then-chief financial officer. Kaplan says he was told that if Overwatch didn't reach certain revenue targets, the publisher would lay off 1,000 people and that would be on Kaplan's shoulders.This article originally appeared on Engadget at https://www.engadget.com/gaming/former-overwatch-director-jeff-kaplan-returns-with-a-western-survival-shooter-161221852.html?src=rss
With Claude enjoying a moment of newfound popularity among regular people, Anthropic is previewing an update designed to make its chatbot better at explaining some concepts. Starting today, Claude can generate charts and diagrams as part of its responses, either when asked directly or when it decides visuals might be helpful to the user. For example, try asking Claude what's the best way to fold a paper plane. Where previously it was limited to text, now it can show you step by step how to fold a Nakamura lock plane. Anthropic is quick to point out what it's introducing today isn't image generation. When producing visual aids, Claude will use HTML code and XML vector graphics. Anthropic likens it to giving Claude access to its own whiteboard. The new feature is available to all Claude users, regardless of whether you pay for one of Anthropic's subscriptions. However, the company does warn it's releasing beta software, so expect some quirks along the way. The feature also isnt available on mobile just yet. This release comes just days after OpenAI made ChatGPT capable of generating interactive visuals when explaining science and math concepts. This article originally appeared on Engadget at https://www.engadget.com/ai/claude-can-now-generate-charts-and-diagrams-160000369.html?src=rss
Honda has announced it is canceling three electric vehicles it was months from starting production on at its EV Hub in Ohio. The Honda 0 SUV, the Honda 0 sedan and the Acura RSX are all being wound down. The company showed off all three models, and touted them as in near-production form at CES 2025. Unlike the Honda Prologue and Acura ZDX, which run on GM's Ultium platform, the scrapped models were built on Honda's own Zero platform and would have been its first fully in-house EVs.
Honda in part blamed the elimination of federal EV tax credits, eased fossil fuel regulations and US tariffs for the decision. The company said US demand for electric vehicles had slowed because of the policy changes. In China, it admitted it could not match the value offered by newer manufacturers building software-driven vehicles on shorter production cycles. CEO Toshihiro Mibe said at a press conference that the demand shift had made EV profitability "very difficult," according to Reuters.
The company said it will redirect resources toward next-generation hybrids and only bring EVs to market when demand justifies it. It also shared plans to expand in India, where it expects the hybrid market to expand. Honda is not alone in pulling back from EV production, with brands like Hyundai, Kia, Volkswagen, Porsche and Ford having all scrapped or delayed major US EV projects since federal policy shifted.
The total restructuring of its EV business could cost Honda up to 2.5 trillion yen ($15.7 billion), with the company set to lose money for the first time since going public in 1957. Mibe and Executive Vice President Noriya Kaihara will forgo 30 percent of their compensation for three months, with other senior executives giving up 20 percent. Honda said it plans to share a detailed long-term strategy at a press conference in May.This article originally appeared on Engadget at https://www.engadget.com/transportation/evs/honda-cancels-three-evs-that-were-months-away-from-us-production-155523384.html?src=rss
Flash floods are notoriously difficult to predict, but Google might have a novel solution. The company just revealed Groundsource, a prediction tool for flash floods that uses Gemini to source data from old news reports. This is the first time it has used a language model for this type of work.
Flash flood prediction models need historical data and model training that often doesn't exist. Our solution: Groundsource, a new AI-powered methodology that uses Gemini to transform 5M+ global reports into a precise dataset of 2.6M+ flood events.This provides a massive, Google Research (@GoogleResearch) March 12, 2026
Google tasked Gemini with sorting through 5 million news articles from around the world and isolating flood reports. It transformed this data into a geo-tagged series of chronological events. Next, researchers trained a model to ingest current weather forecasts and leverage the Groundsource data to determine the likelihood of a flash flood in a given area.
We don't have any concrete information as to how accurate Google's forecast model is, though that should come over time. One trial user did say it helped his organization respond quicker to localized weather events. For now, the company is highlighting risks for urban areas in 150 countries via its Flood Hub platform. Google is also sharing its data with emergency response agencies in these locations.
Google
There are some limitations here. The model can only identify risk across a 20-square-kilometer area. It's also not quite as precise as the US National Weather Service's flood alert system, because Google's model doesn't integrate local radar data. This data typically enables real-time tracking of precipitation. However, the platform's been designed to work in areas that don't typically have access to that kind of weather-sensing infrastructure.
Juliet Rothenberg, a program manager on Google's Resilience team, hopes that this technology can eventually be used to predict other tricky phenomena. This includes stuff like heat waves and mudslides.
"Were aggregating millions of reports, she told reporters this week. "It enables us to extrapolate to other regions where there isnt as much information."
This is Google's first use of a language model for weather forecasts, but not its first time it has relied on AI for this type of thing. The company's DeepMind WeatherNext 2 forecasting model has proven to be extremely accurate.This article originally appeared on Engadget at https://www.engadget.com/ai/google-built-a-flash-flood-prediction-tool-using-gemini-and-old-news-reports-154542963.html?src=rss
