Sensemaking in unfamiliar domains can be challenging, demanding considerable user effort to compare different options with respect to various criteria. Prior research and our formative study found that people would benefit from reading an overview of an information space upfront, including the criteria others previously found useful. However, existing sensemaking tools struggle with the "cold-start" problem -- it not only requires significant input from previous users to generate and share these overviews, but such overviews may also turn out to be biased and incomplete. In this work, we introduce a novel system, Selenite, which leverages Large Language Models (LLMs) as reasoning machines and knowledge retrievers to automatically produce a comprehensive overview of options and criteria to jumpstart users' sensemaking processes. Subsequently, Selenite also adapts as people use it, helping users find, read, and navigate unfamiliar information in a systematic yet personalized manner. Through three studies, we found that Selenite produced accurate and high-quality overviews reliably, significantly accelerated users' information processing, and effectively improved their overall comprehension and sensemaking experience.

Fox News Flash top headlines are here. Check out what's clicking on Foxnews.com. Each morning when she wakes up, Juanita Mengel removes the silicone liner of her prosthetic leg out from under a heated blanket so that the metal parts of the artificial limb don't feel as cold on her skin when she straps the pieces together. The 67-year-old Amanda, Ohio, resident then does the same for her 5-year-old dilute tortoiseshell cat, Lola-Pearl, who is missing her left hind leg. The duo is one of an estimated 200 therapy cat teams registered in the U.S. through Pet Partners.

If you love the deals that come with Amazon Prime Day but don't love Amazon, or don't pay for a Prime membership, you can enjoy similar discounts at Target. The retailer's competing sale, this year called Target Circle Week, runs from July 9 to 15. Amazon's Prime Day falls on July 11 and 12. Nothing beats a free afternoon roaming the aisles of Target, picking up random things as you go along, but some of these discounts are worth the virtual shopping spree. Note: You need to register for Target Circle to see and save the deals. It's free to sign up. You will need to save the offer to your account from the main offer page or on the actual buy page for a particular item to see the deal price at checkout.

If you love the deals that come with Amazon Prime Day but don't love Amazon, or don't pay for a Prime membership, you can enjoy similar discounts at Target. The retailer's competing sale, this year called Target Circle Week, runs from July 9 to 15. Amazon's Prime Day falls on July 11 and 12. Nothing beats a free afternoon roaming the aisles of Target, picking up random things as you go along, but some of these discounts are worth the virtual shopping spree. Note: You need to register for Target Circle to see and save the deals. It's free to sign up. You will need to save the offer to your account from the main offer page or on the actual buy page for a particular item to see the deal price at checkout.

Contemporary autonomous vehicle (AV) benchmarks have advanced techniques for training 3D detectors, particularly on large-scale lidar data. Surprisingly, although semantic class labels naturally follow a long-tailed distribution, contemporary benchmarks focus on only a few common classes (e.g., pedestrian and car) and neglect many rare classes in-the-tail (e.g., debris and stroller). However, AVs must still detect rare classes to ensure safe operation. Moreover, semantic classes are often organized within a hierarchy, e.g., tail classes such as child and construction-worker are arguably subclasses of pedestrian. However, such hierarchical relationships are often ignored, which may lead to misleading estimates of performance and missed opportunities for algorithmic innovation. We address these challenges by formally studying the problem of Long-Tailed 3D Detection (LT3D), which evaluates on all classes, including those in-the-tail. We evaluate and innovate upon popular 3D detection codebases, such as CenterPoint and PointPillars, adapting them for LT3D. We develop hierarchical losses that promote feature sharing across common-vs-rare classes, as well as improved detection metrics that award partial credit to "reasonable" mistakes respecting the hierarchy (e.g., mistaking a child for an adult). Finally, we point out that fine-grained tail class accuracy is particularly improved via multimodal fusion of RGB images with LiDAR; simply put, small fine-grained classes are challenging to identify from sparse (lidar) geometry alone, suggesting that multimodal cues are crucial to long-tailed 3D detection. Our modifications improve accuracy by 5% AP on average for all classes, and dramatically improve AP for rare classes (e.g., stroller AP improves from 3.6 to 31.6)! Our code is available at https://github.com/neeharperi/LT3D

Last fall, I published a book about strollers and what they reveal about our attitudes toward children and their caretakers. Although I pitched Stroller as, in part, a critique of the consumer culture of contemporary American parenthood, I came to love my (many) strollers. In the years when I routinely ran while pushing my kids ahead of me in our jogging stroller, I recorded race times faster than I had as the captain of my college track team. In the long, claustrophobic early days of the pandemic, my son and I meandered slowly up and down the sidewalks of our neighborhood watching that late, cold spring come to New England. Often, at the end of a long stroller walk or run, my kids fell asleep, and on warm days, I'd park them in the shade and myself in the sun to work while they slept, feeling a proud mix of self-sufficiency and frugality (no childcare needed to run or meet a deadline).

A California billionaire has ramped up attacks on Tesla by running a Super Bowl ad questioning the safety of the car maker's self-driving technology. The 30-second commercial shows the electric cars crashing into child-sized mannequins, driving past a stopped school bus and hitting strollers in a parking lot while a narrator proclaims that "Tesla's full self-driving is endangering the public." The ad is the latest in what has been a yearlong campaign by tech executive Dan O'Dowd to have Tesla's Full Self-Driving technology, or FSD, barred from the roads and push lawmakers to increase scrutiny of the technology's safety. Dowd founded a campaign dubbed the Dawn Project to speak out against Tesla, and bugs and security defects in other computer systems. The organization has run a full-page ad in the New York Times and posted similar videos online, but the newest video ran during one of the nation's most watched sporting events, in which a 30-second commercial was reported to cost $6 million to $7 million.

From mundane tools like lawnmowers to fantastical concept cars that get to know their drivers, just about every innovation showcased at CES 2023 was infused with AI. While some products took the form of "robots," others invisibly leveraged AI capabilities to make everyday products smarter and more useful. As AMD CEO Lisa Su said in her keynote address, "AI is truly the most important megatrend for the future of tech." Advancements in autonomous driving haven't developed as quickly as some imagined they would. However, cars are incrementally becoming smarter and smarter, with autonomy seemingly just over the horizon.

This year, the Consumer Electronics Show (CES) -- the celebration of all things tech and then some -- brought what it usually brings to the Vegas Strip: AI-powered gadgets. Of course, AI comes in many forms, and not all that's advertised as AI is in fact true AI. But there's always diamonds in the rough, like an AI-equipped oven from Samsung that automatically adjusts cooking settings for the perfect bake. As my colleagues on the ground traverse the show floor (I'm covering CES remotely this year), they're highlighting the most interesting AI-powered tech they come across. I've compiled each into a handy list, which is far from definitive, but which aims to give an idea of the top AI trends this year at CES. Samsung kicked off CES 2023 with a new oven and fridge in its Bespoke Home smart appliances lineup.

Non-parents may not believe it, but pushing a pram around can be a fairly strenuous task, especially when the train gets rough. It's a full body workout to push two kids under four in my old Uppababy Vista, which weighed the same as an iceberg and had the turning circle of the Titanic. To remedy this, Canadian startup GlüxKind has developed an electrically-assisted stroller that'll make pushing easier, and can even drive itself, albeit only when your kid isn't on board. The GlüxKind Ella is the brainchild of Anne Hunger and Kevin Huang, a couple who were less than whelmed when looking for a stroller for their own daughter. They decided to build their own device by strapping an electric skateboard to a regular stroller, and started developing their product from there. The device has three modes, the first of which is to add electric assist to the wheels as you're pushing it around.