A robotic sub explored a hidden world 1,300 meters under Antarctica. More information Adding us as a Preferred Source in Google by using this link indicates that you would like to see more of our content in Google News results. Breakthroughs, discoveries, and DIY tips sent six days a week. The Antarctic sea, where glaciers drift across the surface. What kind of world lies 1,300 meters below the surface?

The Download: DeepSeek's latest AI breakthrough, and the race to build world models Plus: China has blocked Meta's $2 billion acquisition of AI startup Manus. On Friday, Chinese AI firm DeepSeek released a preview of V4, its long-awaited new flagship model. Notably, the model can process much longer prompts than its last generation, thanks to a new design that handles large amounts of text more efficiently. While the model remains open source, its performance matches leading closed-source rivals from Anthropic, OpenAI, and Google. Here are three ways V4 could shake up AI . AI systems have already gained impressive mastery over the digital world, but the physical world remains humanity's domain.

What are Russia's gains from the Iran war? 'We are not losers; we are winners' Ukrainian officials say Russian drones have again attacked the southern port city of Odesa, injuring at least 11 people, including two children, and damaging homes and important infrastructure. Odesa Governor Oleh Kiper said the attack affected three districts, hitting residential buildings, vehicles and civilian facilities, including a hotel, warehouses and funicular railway. Windows shattered in many buildings and the port area sustained damage. Law enforcement agencies are documenting the latest war crimes committed by Russia against the peaceful population of [the] Odesa region," Kiper said. Russian attacks killed one person in the southeastern Zaporizhzhia region, according to Governor Ivan Fedorov. "A 59-year-old man died as a result of an enemy attack on the Zaporizhzhia region," Fedorov wrote on Telegram. A Ukrainian drone attack killed an employee at the Zaporizhzhia nuclear power plant, which was captured by Russian forces and is shut down. "A driver was killed today when a Ukrainian Armed Forces drone struck the transport department at the Zaporizhzhia Nuclear Power Plant," said a statement from plant managers who were installed by Russia. Regional governor Fedorov said Russian forces launched 629 strikes across 45 settlements in the region in a single day, with at least 50 reports of damage to homes and infrastructure. Russian officials reported Ukrainian drone attacks in the Belgorod border region, where at least one person was killed and four women injured, alongside damage to buildings and vehicles. The attacks come as diplomatic efforts to end the war remain stalled. Donald Trump said on Sunday that he has had "good conversations" with Presidents Vladimir Putin and Volodymyr Zelenskyy. "We're working on the Russia situation, Russia and Ukraine, and hopefully we're going to get it," Trump said on Fox News. "I do have conversations with him, and I do have conversations with President Zelenskyy, and good conversations," he said. "The hatred between President Putin and President Zelenskyy is ridiculous.

Datacentres could require at least 6GW of capacity by 2030 under government plans to expand AI infrastructure. Datacentres could require at least 6GW of capacity by 2030 under government plans to expand AI infrastructure. Sun 26 Apr 2026 03.00 EDTLast modified on Sun 26 Apr 2026 03.01 EDT One vision of the UKâ s future involves a decarbonised economy powered by clean, renewable energy. Another involves making the UK an AI superpower. The government departments responsible for these two visions do not appear to have agreed on their numbers.

Silicon-photonics-based optical neural network (ONN) is a promising hardware platform that could represent a paradigm shift in efficient AI with its CMOScompatibility, flexibility, ultra-low execution latency, and high energy efficiency. In-situ training on the online programmable photonic chips is appealing but still encounters challenging issues in on-chip implementability, scalability, and efficiency. In this work, we propose a closed-loop ONN on-chip learning framework L2ight to enable scalable ONN mapping and efficient in-situ learning. L2ightadopts a three-stage learning flow that first calibrates the complicated photonic circuit states under challenging physical constraints, then performs photonic core mapping via combined analytical solving and zeroth-order optimization. A subspace learning procedure with multi-level sparsity is integrated into L2ightto enable in-situ gradient evaluation and fast adaptation, unleashing the power of optics for real on-chip intelligence. Extensive experiments demonstrate our proposed L2ightoutperforms prior ONN training protocols with 3-order-of-magnitude higher scalability and over 30 better efficiency, when benchmarked on various models and learning tasks. This synergistic framework is the first scalable on-chip learning solution that pushes this emerging field from intractable to scalable and further to efficient for next-generation self-learnable photonic neural chips. From a co-design perspective, L2ightalso provides essential insights for hardware-restricted unitary subspace optimization and efficient sparse training.

Bayesian optimization (BO) conventionally relies on handcrafted acquisition functions (AFs) to sequentially determine the sample points. However, it has been widely observed in practice that the best-performing AF in terms of regret can vary significantly under different types of black-box functions. It has remained a challenge to design one AF that can attain the best performance over a wide variety of black-box functions. This paper aims to attack this challenge through the perspective of reinforced few-shot AF learning (FSAF). Specifically, we first connect the notion of AFs with Q-functions and view a deep Q-network (DQN) as a surrogate differentiable AF. While it serves as a natural idea to combine DQN and an existing few-shot learning method, we identify that such a direct combination does not perform well due to severe overfitting, which is particularly critical in BO due to the need of a versatile sampling policy. To address this, we present a Bayesian variant of DQN with the following three features: (i) It learns a distribution of Q-networks as AFs based on the Kullback-Leibler regularization framework. This inherently provides the uncertainty required in sampling for BO and mitigates overfitting.