This week’s great tech stories from around the web (through April 9th)


OpenAI’s DALL-E 2 produces fantastic visuals of almost anything you can imagine
Andrew Tarantola | Engadget
“DALL-E 2, which leverages OpenAI’s CLIP image recognition system, builds on these image generation capabilities. Users can now select and edit specific parts of existing images, add or remove elements and their shadows, merge two images into a single collage, and generate variations of an existing image. In addition, the output images are 1024px squares, an increase from the 256px avatars generated by the original version. OpenAI’s CLIP is designed to look at a particular image and summarize its content in a way that people can understand. The consortium reversed that process and built a picture of the summary in its work with the new system.”


NASA sets new date for first Launchpad test of its Mega Moon rocket
Trevor Mogg | Digital trends
“If everything goes according to plan and there are no more technical problems, NASA will hopefully be able to prepare for the first launch of the SLS rocket and the Orion spacecraft in the coming months. The unmanned Artemis I mission will send Orion on a flight around the moon for an extensive test of its spaceflight systems. Artemis II will fly the same route, but with a crew on board, while the highly anticipated Artemis III mission, currently scheduled for no earlier than 2024, will bring the first woman and first person of color to the lunar surface.


While Russia plots its next move, an AI listens to the chatter
Will Knight | wired
“While the soldiers spoke, an AI listened. Their words were automatically captured, transcribed, translated and analyzed using various artificial intelligence algorithms developed by Primer, an American company that provides AI services to intelligence analysts. While it is not clear whether Ukrainian forces also intercepted the communications, the use of AI systems to conduct large-scale surveillance of the Russian military shows the growing importance of advanced open source intelligence in military conflicts. A number of unsecured Russian broadcasts have been posted online, translated and analyzed on social media. Other data sources, including video clips from smartphones and social media posts, were similarly examined. But using natural language processing technology to analyze Russian military communications is very new.”


Elon Musk is on the board of Twitter. What could go wrong?
Chris Stokel-Walker | wired
Musk’s hot and cold relationship with Twitter sheds little light on why he bought into the company and joined its board of directors, although theories abound. Musk did not respond to a request for comment. One hint was too found in his recent Twitter posts.The entrepreneur has long been an open book on the social network, saying in 2018 that “my tweets are literally what I think right now, not carefully crafted corporate bs, which is really just banal propaganda.” .’ And recent tweets have focused on the future direction of the platform.Since buying his stake in the company, Musk has questioned his followers about whether Twitter should open its algorithm to scrutiny and whether the platform adheres to the principle of freedom. of speech.”


This Japanese robot’s latest skill: peeling a banana
University of Tokyo via Reuters | NBC News
“While the double-arm machine is only successful 57 percent of the time, peeling bananas points to a future where machines perform more subtle operations than moving metal parts or delivering coffee. Video by researchers at the University of Tokyo shows the robot picking up a banana and peeling it with both hands in about three minutes. Researchers Heecheol Kim, Yoshiyuki Ohmura and Yasuo Kuniyoshi trained the robot using a “deep imitation learning” process where they demonstrated the banana peel action hundreds of times to produce enough data for the robot to learn and replicate its actions.”


Nvidia’s Next GPU Shows Transformers Transforming AI
Samuel K. Moore | IEEE spectrum
“The secret sauce of the transformer motor is its ability to dynamically choose the precision needed for each layer in the neural network at each step in training a neural network. The least accurate units, the 8-bit floating point, can speed up their calculations, but then produce 16-bit or 32-bit sums for the next layer if that’s the precision needed there. However, Hopper goes one step further. The 8-bit floating point units can perform their matrix calculation with one of two forms of 8-bit numbers.”


Tesla officially opens Texas Gigafactory
Will Douglas Heaven | MIT Technology Review
“It is the company’s fourth plant in the US, after the car factory in Fremont, California, the battery factory in Sparks, Nevada and the solar factory in Buffalo, New York. Tesla also has an auto factory outside of Shanghai, China, and recently opened its first European factory near Berlin, Germany. Tesla spent an estimated $5 million to purchase the land outside of Austin, plus another $1.1 billion to build the plant. “We need a place where we can be really big, and there’s no place like Texas,” Musk said. ‘We’re going to a really huge scale.’”

Image Credit: tunnelmotions / 71 images

Leave a Reply

Your email address will not be published.