Generative AI Creates Playable Version of Doom Game With No Code
Matthew Sparkes | New Scientist
“An AI-generated re-creation of the basic pc sport Doom may be performed usually regardless of having no pc code or graphics. Researchers behind the undertaking say related AI fashions could possibly be used to create video games from scratch sooner or later, simply as they create textual content and pictures at this time.”
Robot Metalsmiths Are Resurrecting Toroidal Tanks for NASA
Evan Ackerman | IEEE Spectrum
“Due to their comparatively complicated form, toroidal tanks are rather more tough to make than spherical tanks. Despite the fact that these tanks can carry out higher, NASA merely doesn’t have the experience to fabricate them anymore, since every one needs to be hand-built by extremely expert people. However an organization known as Machina Labs thinks that they’ll do that with robots as an alternative. And their imaginative and prescient is to fully change how we make issues out of metallic.”
The Year Is 2149 and…
Sean Michaels | MIT Expertise Assessment
“Novelist Sean Michaels envisions what life will appear to be 125 years from now: ‘The yr is 2149 and folks largely reside their lives “on rails.” That’s what they name it, “on rails,” which is to reside in keeping with the meticulous directions of software program. Software program is aware of most issues about you—what causes you anxiousness, what raises your endorphin ranges, every part you’ve ever looked for, in all places you’ve been. Software program sends messages in your behalf; it listens in on conversations. ‘”
OpenAI in Talks for Funding Round Valuing It Above $100 Billion
Tom Dotan and Berber Jin | The Wall Road Journal
“The brand new funding spherical could be the largest infusion of outdoor capital into OpenAI since Microsoft invested round $10 billion in January 2023. Since then, an arms race has developed in Silicon Valley to construct probably the most superior artificial-intelligence programs in an effort to dominate an trade many say will revolutionize the financial system. OpenAI was final valued at $86 billion late final yr, when staff offered present shares.”
AI Inference Competition Heats Up
Dina Genkina | IEEE Spectrum
“Whereas the dominance of Nvidia GPUs for AI coaching stays undisputed, we could also be seeing early indicators that, for AI inference, the competitors is gaining on the tech large, significantly when it comes to energy effectivity. The sheer efficiency of Nvidia’s new Blackwell chip, nevertheless, could also be arduous to beat.”
World’s Largest Sailing Cargo Ship Makes First Transatlantic Voyage
Jeremy Hsu | New Scientist
“The world’s largest crusing cargo ship is making its maiden voyage throughout the Atlantic Ocean. It left a port in France in early August, and it’s on monitor to ship 1,000 tons of cognac and champagne to New York Metropolis by 3 September. Its shipments have a carbon footprint one tenth that of an ordinary container ship.”
Thought-to-Text Chip Smaller Than Neuralink Achieves 91% Accuracy
Michael Franco | New Atlas
“The chip has been developed by researchers at [EPFL] and represents a leap ahead within the scorching area of brain-machine interfaces (BMIs)—gadgets which are capable of learn exercise within the mind and translate it into real-world output akin to textual content on a display screen. That’s as a result of this explicit gadget—generally known as a miniaturized brain-machine interface (MiBMI)—is extraordinarily small, consisting of two skinny chips measuring simply 8 mm2 whole.”
We’ll Soon Get the Sharpest Image Ever of a Black Hole
Isaac Schultz | Gizmodo
“Checks by the Occasion Horizon Telescope Collaboration have yielded the highest-resolution observations ever obtained from Earth, laying an thrilling basis for future observations of black holes. …Sometimes, astronomers will get greater decision pictures through the use of an even bigger telescope, however the EHT already spans the Earth. As an alternative, the collaboration noticed shorter wavelengths of sunshine, yielding sharper pictures.”
OpenAI Searches for an Answer to Its Copyright Problems
Elizabeth Lopatto | The Verge
“The large leaps in OpenAI’s GPT mannequin most likely got here from sucking down your complete written internet. That features whole archives of main publishers akin to Axel Springer, Condé Nast, and The Related Press—with out their permission. However for some purpose, OpenAI has introduced offers with many of those conglomerates anyway. At first look, this doesn’t solely make sense. Why would OpenAI pay for one thing it already had? And why would publishers, a few of whom are lawsuit-style offended about their work being stolen, agree?”
DIGITAL MEDIA
Chatbots Are Primed to Warp Reality
Matteo Wong | The Atlantic
“A large physique of analysis, alongside conversations I’ve not too long ago had with a number of specialists, means that the solicitous, authoritative tone that AI fashions take—mixed with them being legitimately useful and proper in lots of circumstances—could lead on individuals to position an excessive amount of belief within the know-how. That credulity, in flip, might make chatbots a very efficient instrument for anybody in search of to control the general public via the refined unfold of deceptive or slanted info.”
Picture Credit score: Nicolas Houdayer / Unsplash
