Decomposing the dark matter of sparse autoencoders.
Joshua Engels, Logan Riggs, Max Tegmark MIT 2024 https://arxiv.org/abs/2410.
On mapping concepts in artificial neural networks with sparse autoencoders: we find that map errors exhibit…
Code for our paper ‘Decomposing The Dark Matter of Sparse Autoencoders’ — JoshEngels/SAE-Dark-Matter.
ROBOT ACHIEVEMENT DAY! 
https://finance.yahoo.com/news/kepler-debuts-forerunner-k2-h…00317.html
SHANGHAI, Oct. 21, 2024 /PRNewswire/ — Shanghai Kepler Robotics Co., Ltd. (“Kepler Humanoid Robot”) has recently launched its full-sized, general-purpose humanoid robot, the Forerunner K2, at GITEX GLOBAL 2024, which began on October 14.
Kepler Humanoid Robot is dedicated to transforming productivity through cutting-edge technology and delivering industry-leading, high-IQ blue-collar humanoid robots. Hu Debo, CEO of Kepler Humanoid Robot, said, “The Forerunner K2 represents the Gen 5.0 robot model, showcasing a seamless integration of the humanoid robot’s cerebral, cerebellar, and high-load body functions. At Kepler, we understand that innovation is driven by application. That’s why we prioritize deep integration of customer needs, solutions, and product development. From the outset, we have forged close collaborations with key customers, jointly formulating commercialization strategies and working alongside industry stakeholders to accelerate the deployment of humanoid robots.”
Posted in robotics/AI
Optica l metasurfaces, planar artificial media capable of controlling light propagation, areioning from laboratory curiosity to commercial applications.
npj Nanophotonics volume 1, Article number: 36 (2024) Cite this article.
When I was a kid we had the anarchist cookbook.
But an artist and hacker found a way to trick ChatGPT to ignore its own guidelines and ethical responsibilities to produce instructions for making powerful explosives.
The hacker, who goes by Amadon, called his findings a “social engineering hack to completely break all the guardrails around ChatGPT’s output.” An explosives expert who reviewed the chatbot’s output told TechCrunch that the resulting instructions could be used to make a detonatable product and was too sensitive to be released.
Analyze the NVIDIA Container Toolkit CVE-2024–0132 vulnerability and its implications for AI system security with recommended remediation steps.
Amazon is using a new, proprietary AI solution called Vision-Assisted Package Retrieval (VAPR) to reduce the time and effort it takes for delivery drivers to locate packages in their vans. Is it a game-changer, or more AI VAPR-ware?
Somewhat lost in the hype surrounding the Tesla “We, Robot” event on 10/10 was an AI press release from Amazon that promises to improve the lives of the people who use it today, rather than two years from now.
VAPR makes Amazon delivery drivers’ lives easier by automatically identifying the packages to be delivered at a given stop. The system then project a green “O” on all packages that will be delivered at that stop, and a red “X” on all other packages. When the driver picks up all the “correct” packages, VAPR delivers an audible cue to help ensure no packages get left behind.
AI and politics 
Artificial though it may be, the concept of “intelligence” doesn’t seem to jibe with a computer-generated image of uniformed cats toting assault rifles.
Yet that visual slur, which supports a debunked story about immigrants in Ohio eating pets, has become a signature image from…
UMD experts explain the emotional pulls and cognitive pitfalls—and how to avoid them.
In a world powered by artificial intelligence applications, data is king, but it’s also the crown’s biggest burden.
As described in the article, quantum memory stores data in ways that classical memory systems cannot match. In quantum systems, information is stored in quantum states, using the principles of superposition and entanglement to represent data more efficiently. This ability allows quantum systems to process and store vastly more information, potentially impacting data-heavy industries like AI.
In a 2021 study from the California Institute of Technology, researchers showed that quantum memory could dramatically reduce the number of steps needed to model complex systems. Their method proved that quantum algorithms using memory could require exponentially fewer steps, cutting down on both time and energy. However, this early work required vast amounts of quantum memory—an obstacle that could have limited its practical application.
Now, two independent teams have derived additional insights, demonstrating how these exponential advantages can be achieved with much less quantum memory. Sitan Chen from Harvard University, along with his team, found that just two quantum copies of a system were enough to provide the same computational efficiency previously thought to require many more.
Have you ever wanted to travel through time to see what your future self might be like?
The user engages with the tool in two ways: through introspection, when they consider their life and goals as they construct their future selves, and retrospection, when they contemplate whether the simulation reflects who they see themselves becoming, says Yin.
“You can imagine Future You as a story search space. You have a chance to hear how some of your experiences, which may still be emotionally charged for you now, could be metabolized over the course of time,” she says.
To help people visualize their future selves, the system generates an age-progressed photo of the user. The chatbot is also designed to provide vivid answers using phrases like “when I was your age,” so the simulation feels more like an actual future version of the individual.
