You don't need to worry about recursive-self-improving AI - yet One of the world's leading artificial intelligence companies has implored the industry to pause development on AI, because the latest models could be reaching a tipping point where they become capable of redesigning themselves, growing ever more powerful and finally escaping our control. At least, that's what the headlines said. In truth, Anthropic's co-founder Jack Clark and the boss of spin-out think-tank The Anthropic Institute, Marina Favaro, have published a long blog post bigging up the capabilities of their Claude model, shortly before the company floats on the stock exchange in an initial public offering (IPO) for a rumoured $1 trillion. Let's, for a moment, ignore the vast financial elephant in the room and look at the technological claims. An AI that becomes capable of designing a more powerful version of itself, which is in turn able to pull off the same feat, is an obvious gamechanger, but it is also not a new idea.

Metalenz's Polar ID face-scanning technology works even when the camera is hidden under the display. The notch has largely been replaced on today's smartphones by floating punch-hole cameras that take up less space and look a little more futuristic, though notches are still prevalent on some laptops, like Apple's MacBooks . On the iPhone, Apple calls its floating pill-shaped camera system the Dynamic Island, which debuted on the iPhone 14 . The iPhone still has the largest camera cutout today, due to its Face ID biometric authentication system. This island could get much smaller, however, thanks to new under-display camera technology announced at Display Week 2026 from Metalenz, a optics startup from Boston.

Obsessed with throwing money and resources at AI in any way they can, the likes of OpenAI, NVIDIA, Google and Amazon all just got a surprise. Out of seemingly nowhere, Chinese AI assistant DeepSeek is suddenly the top-rated free app on Apple's App Store in the US and elsewhere, beating more familiar names, like ChatGPT. The open-source DeepSeek V3 model reportedly requires far less computing power than its competitors and, depending on who you believe, was developed for under 6 million. Shocks all around -- especially for OpenAI and all the billions it has floating around. Focusing on coding and research, DeepSeek's models are similar to other AI assistants you've heard of.

Synthetic active collectives, composed of many nonliving individuals capable of cooperative changes in group shape and dynamics, hold promise for practical applications and for the elucidation of guiding principles of natural collectives. However, the design of collective robotic systems that operate effectively without intelligence or complex control at either the individual or group level is challenging. We investigate how simple steric interaction constraints between active individuals produce a versatile active system with promising functionality. Here we introduce the link-bot: a V-shape-based, single-stranded chain composed of active bots whose dynamics are defined by its geometric link constraints, allowing it to possess scale- and processing-free programmable collective behaviors. A variety of emergent properties arise from this dynamic system, including locomotion, navigation, transportation, and competitive or cooperative interactions. Through the control of a few link parameters, link-bots show rich usefulness by performing a variety of divergent tasks, including traversing or obstructing narrow spaces, passing by or enclosing objects, and propelling loads in both forward and backward directions. The reconfigurable nature of the link-bot suggests that our approach may significantly contribute to the development of programmable soft robotic systems with minimal information and materials at any scale.

We propose an interpretable framework for reading analog gauges that is deployable on real world robotic systems. Our framework splits the reading task into distinct steps, such that we can detect potential failures at each step. Our system needs no prior knowledge of the type of gauge or the range of the scale and is able to extract the units used. We show that our gauge reading algorithm is able to extract readings with a relative reading error of less than 2%.

If you're not interested in a midrange smartphone but don't want to pay flagship prices either, Qualcomm has a possible solution. The company just unveiled the Snapdragon 8s Gen 3 processor that offers flagship features, but performance just below the Snapdragon 8 Gen 3. It's built using a 4-nanometer (nm) process with a 1 4 3 core setup (one prime ARM Cortex X4 at 3.0GHz, four performance cores at 2.8Ghz and three 2.0GHz efficiency cores. That compares with the Snapdragon 8 Gen 3's 1 5 2 setup, so the "s" model swaps a performance core for an efficiency core. The prime core on the flagship chip also runs a bit faster at 3.4Ghz. The chip uses a previous-gen X70 5G modem, though it still offers Wi-Fi 7 support.

Value alignment is essential for building AI systems that can safely and reliably interact with people. However, what a person values--and is even capable of valuing--depends on the concepts that they are currently using to understand and evaluate what happens in the world. The dependence of values on concepts means that concept alignment is a prerequisite for value alignment--agents need to align their representation of a situation with that of humans in order to successfully align their values. Here, we formally analyze the concept alignment problem in the inverse reinforcement learning setting, show how neglecting concept alignment can lead to systematic value mis-alignment, and describe an approach that helps minimize such failure modes by jointly reasoning about a person's concepts and values. Additionally, we report experimental results with human participants showing that humans reason about the concepts used by an agent when acting intentionally, in line with our joint reasoning model. People's thoughts and actions are fundamentally shaped by the concepts they use to represent the world and formulate their goals.

Summer is almost here, and if you have some cool adventures planned, you're going to want to record them for posterity. With a drone, you can take your Instagram feed up a notch, and during our Memorial Day Sale, you can get two drones for less than the price of one. The Alpha Z PRO 4K Flying Fox 4K Wide-Angle Dual-Camera Drone Bundle includes an Alpha Z PRO Ultra HD Dual-Camera Drone and a Flying Fox Ultra HD Dual-Camera Drone for one price. Both drones have a 4K HD front camera and a 720p bottom camera, as well as altitude hold modes, 4-channel movement, and real-time FPV for an easier and more immersive flight experience. They're both headless drones with one-key automatic return so they're suitable even for beginners to fly.

"Ye shall know them by the temporal dynamics of their temperature parameter." God created the universe by strategically forgetting about something that would take a whole reality to remember. It doesn't mean He (gender neutral) did it on purpose. Everything that can happen, happens. See, the Hindus thought that the "way out" was by "remembering who you are".

TL;DR -- Someone entered an art competition with an AI-generated piece and won the first prize.Yeah that's pretty… https://t.co/SFtD8Oas7M It's almost like artists are smart people and can tell you exactly how this kind of thing will devalue what we do and you should listen to us when we point out the problems. We're watching the death of artistry unfold right before our eyes -- if creative jobs aren't safe from machines, then even high-skilled jobs are in danger of becoming obsolete What will we have then? I've met plenty of "idea guys" that just *wish* they could draw. In some aspects I like what one could use AI images for, but I refuse to accept it as "art" in the same vien as something created directly by a person.