For decades, Cai Guo-Qiang has been the world's foremost fine artist of explosions. He is famous for his massive fireworks displays, from his glowing footsteps in the sky at the opening of the 2008 Beijing Olympics, to his 2015 Sky Ladder, a 1,650-foot flaming ladder to heaven featured in a Netflix documentary. Recently, the gunpowder artist has become obsessed with a new threatening technology: artificial intelligence. AI "brings me more anxiety, but also, freshness", the 66-year-old Chinese artist told me last week at the historic Nassau Veterans Memorial Coliseum in Los Angeles, where he was preparing for his newest "explosion event", which would be the kickoff of a major arts festival opening in southern California this month. "It's similar to why I use gunpowder," Cai told me.

A monster AI machine with blades is revolutionizing the world of lawn maintenance. Don't yawn and click off the page just because this story begins with grass. Yes, the stuff that grows in your backyard. You see, the world of lawn care just got its equivalent of a supersonic jet -- the autonomous mega-mower by FireFly Automatix. This is the mega-mower, a monster that'll blitz an area equivalent to 19 NFL football fields on a single charge.

Recent research has focused on enhancing the capability of smaller models through imitation learning, drawing on the outputs generated by large foundation models (LFMs). A number of issues impact the quality of these models, ranging from limited imitation signals from shallow LFM outputs; small scale homogeneous training data; and most notably a lack of rigorous evaluation resulting in overestimating the small model's capability as they tend to learn to imitate the style, but not the reasoning process of LFMs. To address these challenges, we develop Orca (We are working with our legal team to publicly release a diff of the model weights in accordance with LLaMA's release policy to be published at https://aka.ms/orca-lm), a 13-billion parameter model that learns to imitate the reasoning process of LFMs. Orca learns from rich signals from GPT-4 including explanation traces; step-by-step thought processes; and other complex instructions, guided by teacher assistance from ChatGPT. To promote this progressive learning, we tap into large-scale and diverse imitation data with judicious sampling and selection. Orca surpasses conventional state-of-the-art instruction-tuned models such as Vicuna-13B by more than 100% in complex zero-shot reasoning benchmarks like Big-Bench Hard (BBH) and 42% on AGIEval. Moreover, Orca reaches parity with ChatGPT on the BBH benchmark and shows competitive performance (4 pts gap with optimized system message) in professional and academic examinations like the SAT, LSAT, GRE, and GMAT, both in zero-shot settings without CoT; while trailing behind GPT-4. Our research indicates that learning from step-by-step explanations, whether these are generated by humans or more advanced AI models, is a promising direction to improve model capabilities and skills.

Incredible drone footage has captured video of the largest ever gathering of sea turtles near a wildlife refuge in Costa Rica as they prepare to lay their eggs. Every year hundreds of thousands of female sea turtles arrive at the Ostional Nacional Wildlife Refuge within a few days of each other. Biologist Vanessa Bézy has been studying this phenomenon - called an arribada - and filmed olive ridley sea turtles aggregating in the ocean before reaching shore. She said she saw thousands of turtles per square mile during the filming. Sea turtles are at their most vulnerable to predators when they are tiny hatchlings on the beach and so large groups of mothers lay their eggs at the same time in the same area to increase the chance of their offspring surviving.

The 23-year-old wearing a sweatshirt and funky jogger pants could almost be mistaken for any other young man engaging in lighthearted banter while playing video games on a Tuesday morning. But after an MVP award in his second NFL season, Patrick Mahomes II tends to stick out, even when engaging in everyday activities like gaming. Though so much has changed in Mahomes's life after he laid waste to NFL secondaries with more than 5,000 yards and 50 touchdowns, he still reverts to a ritual that dates back to his middle school days: hours-long offseason video gaming sessions. "It's been something that I've taken with me all the way since I was that young," Mahomes said during a promotional event for new in-game content for "Call of Duty: Black Ops 4" at Treyarch's offices. Video games, especially shooters like Call of Duty, have served as a way for Mahomes to escape the pressures of his life on the field while still flexing his competitive muscles.

The sensor technology that helps athletes avoid injury and improve performance can bring benefits to the enterprise. Professional football in the United States is reaching a fever pitch, with the playoffs concluded and fans looking to the Super Bowl next month. When you hear the crush of helmets while you're watching, think sensors. On the football field, in the enterprise, and beyond, organizations are using sensors to keep employees safe, to enhance their performance, and to improve employee satisfaction. For example, a team doctor could know that a player has a concussion before the player even hits the ground.

A set diagram depicts the places where different groups of objects overlap. Example: in the United States, there are about 1,700 professional football players, and thousands of people pursuing PhDs in math. On an overcast day in late winter, that place is the Norbert Wiener Common Room in MIT's Department of Mathematics, where John Urschel is sitting at a table, chatting. Urschel is an offensive lineman with the NFL's Baltimore Ravens, a three-year pro with 40 regular- season games played and a couple of playoff starts on his football résumé. He is also a doctoral candidate in math at MIT who has passed his qualifying exams and has nine published or accepted research papers on his academic résumé.

A set diagram depicts the places where different groups of objects overlap. Example: in the United States, there are about 1,700 professional football players, and thousands of people pursuing PhDs in math. On an overcast day in late winter, that place is the Norbert Wiener Common Room in MIT's Department of Mathematics, where John Urschel is sitting at a table, chatting. Urschel is an offensive lineman with the NFL's Baltimore Ravens, a three-year pro with 40 regular- season games played and a couple of playoff starts on his football résumé. He is also a doctoral candidate in math at MIT who has passed his qualifying exams and has nine published or accepted research papers on his academic résumé.

Ford has unveiled a sleeker design for its self-driving Fusion. The new model is still equipped with the current autonomous vehicle platform, but now has more computing power that will help the firm unleash them on the road by 2021. Also, unlike other autonomous models, the American car maker removed the bundle of sensors from the roof, giving this prototype a more traditional automobile look. Ford has unveiled a sleeker design of it self-driving car Fusion. It process data transmitted from an array of sensors that help the car'see' its surroundings.

From 1992 to 2001, Dan Goldin served as the longest-tenured administrator of NASA, overseeing projects like the launch of the Space Shuttle Endeavor and the redesign of the International Space Station. After leaving NASA, Goldin spent some time bouncing around and studying robotics before accepting a position as the president of Boston University in 2003. He never officially held the position, however, because the school terminated his contract a day before he was slated to start (though he still got a 1.8 million payout). And then Goldin mostly vanished from the public eye for over 10 years. Today, the 75-year-old Goldin has reemerged to reveal what he has been working on for the past decade: KnuEdge, a top-secret startup based in San Diego, with a mission to one-up Google, AMD, and Intel with the "fundamental invention" of the next-generation computer processor.