Latest news with #Asimov
WIRED
16-07-2025
- Business
- WIRED
Former Top Google Researchers Have Made A New Kind of AI Agent
Jul 16, 2025 9:02 AM The mission? Teaching models to better understand how to build code will lead to superintelligent AI. Photo-Illustration:A new kind of artificial intelligence agent, trained to understand how software is built by gorging on a company's data and learning how this leads to an end product, could be both a more capable software assistant and a small step towards much smarter AI. The new agent, called Asimov, was developed by Reflection, a small but ambitious startup confounded by top AI researchers from Google. Asimov reads code as well as emails, Slack messages, project updates and other documentation with the goal of learning how all this leads together to produce a finished piece of software. Reflection's ultimate goal is building superintelligent AI—something that other leading AI labs say they are working towards. Meta recently created a new Superintelligence Lab, promising huge sums to researchers interested in joining its new effort. I visited Reflection's headquarters in the Brooklyn neighborhood of Williamsburg, New York, just across the road from a swanky-looking pickleball club, to see how Reflection plans to reach superintelligence ahead of the competition. The company's CEO, Misha Laskin, says the ideal way to build supersmart AI agents is to have them truly master coding, since this is the simplest, most natural way for them to interact with the world. While other companies are building agents that use human user interfaces and browse the web, Laskin, who previously worked on Gemini and agents at Google DeepMind, says this hardly comes naturally to a large language model. Laskin adds that teaching AI to make sense of software development will also produce much more useful coding assistants. Laskin says Asimov is designed to spend more time reading code rather than writing it. 'Everyone is really focusing on code generation,' he told me. 'But how to make agents useful in a team setting is really not solved. We are in kind of this semi-autonomous phase where agents are just starting to work.' Asimov actually consists of several smaller agents inside a trench coat. The agents all work together to understand code and answer users' queries about it. The smaller agents retrieve information, and one larger reasoning agent synthesizes this information into a coherent answer to a query. Reflection claims that Asimov already is perceived to outperform some leading AI tools by some measures. In a survey conducted by Reflection, the company found that developers working on large open source projects who asked questions preferred answers from Asimov 82 percent of the time compared to 63 percent for Anthropic's Claude Code running its model Sonnet 4. Daniel Jackson, a computer scientist at Massachusetts Institute of Technology, says Reflection's approach seems promising given the broader scope of its information gathering. Jackson adds, however, that the benefits of the approach remain to be seen, and the company's survey is not enough to convince him of broad benefits. He notes that the approach could also increase computation costs and potentially create new security issues. 'It would be reading all these private messages,' he says. Reflection says the multiagent approach mitigates computation costs and that it makes use of a secure environment that provides more security than some conventional SaaS tools. In New York, I met with the startup's CTO, Ioannis Antonoglou. His expertise training AI models to reason and play games is being applied to having them build code and do other useful chores. A founding engineer at Google DeepMind, Antonoglou did groundbreaking research on a technique known as reinforcement learning, which was most famously used to build AlphaGo, a program that learned to play the ancient board game Go to a superhuman level using . Reinforcement learning, which involves training an AI model through practice combined with positive and negative feedback, has come to the fore in the past few years because it provides a way to train a large language model to produce better outputs. Combined with human training, reinforcement learning can train an LLM to provide more coherent and pleasing answers to queries. With additional training, reinforcement learning helps a model learn to perform a kind of simulated reasoning, whereby tricky problems are broken into steps so that they can be tackled more effectively. Asimov currently uses open source models but Reflection is using reinforcement learning to post-train custom models that it says perform even better. Rather than learning to win at a game like Go, the model learns how to build a finished piece of software. Tapping into more data across a company should . Reflection uses data from human annotators and also generates its own synthetic data. It does not train on data from customers. Big AI companies are already using reinforcement learning to tune agents. An OpenAI tool called Deep Research, for instance, uses feedback from expert humans as a reinforcement learning signal that teaches an agent to comb through websites, hunting for information on a topic, before generating a detailed report. 'We've actually built something like Deep Research but for your engineering systems,' Antonoglou says, noting that training on more than just code provides an edge. 'We've seen that in big engineering teams, a lot of the knowledge is actually stored outside of the codebase.' Stephanie Zhan, a partner at the investment firm Sequoia, which is backing Reflection, says the startup 'punches at the same level as the frontier labs.' With the AI industry now shooting for superintelligence, and deep pocketed companies like Meta pouring huge sums into hiring and building infrastructure, startups like Reflection may find it more challenging to compete. I asked Reflection leaders what the path to more advanced might actually look like. They believe an increasingly intelligent agent would go on to become an oracle for companies' institutional and organizational knowledge. It should learn to build and repair software autonomously. Eventually it would invent new algorithms, hardware, and products autonomously. The most immediate next step might be less grand. 'We've actually been talking to customers who've started asking, can our technical sales staff, or our technical support team use this?' Laskin says.
New Indian Express
14-07-2025
- Politics
- New Indian Express
Risks and realities of killer robots
In his sci-fi novel, Runaround, Isaac Asimov introduced the three laws of robotics to explore the moral boundaries of machine intelligence. His robots were programmed to preserve human life, obey ethical constraints, and act only within a tightly defined moral architecture. These laws forced readers to grapple with the limits of delegation and the necessity of conscience in decision-making. This insight is especially relevant today, as warfare increasingly incorporates unmanned systems. In recent conflicts—India's Operation Sindoor, Azerbaijan's use of Turkish drones against Armenian forces, and Ukraine's deep drone strikes into Russian territory— all offensive systems remained human-operated. Humans directed target selection, authorisation and engagement. But now, as the global defence landscape shifts toward lethal autonomous weapon systems (LAWS), Asimov's warning grows more relevant. Unlike the author's fictional robot Speedy, these systems will not hesitate when ethical ambiguities arise. They will not wait for human correction. They will act without the possibility of a moral pause. LAWS are weapons that can select, track, and engage targets without real-time human control. LAWS rely on AI, sensor fusion, and machine learning algorithms to make independent targeting decisions. This autonomy dramatically accelerates response time and expands operational reach, but at significant ethical and legal cost. The development of LAWS is already underway in multiple countries. The US, China, Russia, Israel and South Korea have invested heavily in autonomous platforms ranging from loitering munitions to swarming drones and autonomous ground systems. The US military has demonstrated autonomous swarms in exercises like Project Convergence; China is integrating AI into hypersonic systems and naval platforms; and Russia has tested autonomous tanks like Uran-9. Although fully autonomous systems capable of making unsupervised kill decisions are not yet officially deployed, the technological threshold is narrowing.
Gulf Insider
14-07-2025
- Science
- Gulf Insider
China Warns of Rogue Robot Troops Unleashing
Concerns are mounting in China as the Communist superpower advances humanoid robot development to replace human soldiers on the battlefield, prompting calls for 'ethical and legal research' into this Terminator-like technology to 'avoid moral pitfalls.' An op-ed published by Yuan Yi, Ma Ye and Yue Shiguang in the People's Liberation Army (PLA) Daily warned that faulty robots could lead to 'indiscriminate killings and accidental death,' which would 'inevitably result in legal charges and moral condemnation.' The South China Morning Post reports: The authors cited American science fiction writer Isaac Asimov's Three Laws of Robotics, a set of principles that have influenced discussions about the ethics of real-world applications in the field. The authors said that militarised humanoid robots 'clearly violate' the first of Asimov's laws, which states that a robot 'may not injure a human being or, through inaction, allow a human being to come to harm'. They added that Asimov's laws needed to be overhauled in the light of these developments. They also highlighted legal implications, saying that humanoid robots in military scenarios should comply with the main principles of the laws of war by 'obeying humans', 'respecting humans' and 'protecting humans'. The authors emphasized that robots must be designed with constraints to 'suspend and limit excessive use of force in a timely manner and not indiscriminately kill people.' Additionally, the trio cautioned against hastily replacing humans with robots, noting that robots still lack essential capabilities such as speed, dexterity, and the ability to navigate complex terrains. 'Even if humanoid robots become mature and widely used in the future, they will not completely replace other unmanned systems,' the article said. Concurrently, the U.S. Army is intensifying efforts to integrate robotics, artificial intelligence, and autonomous systems, aiming to enhance human-machine collaboration between soldiers and advanced robots on the battlefield, according to Interesting Engineering. Scientists at the U.S. Army Combat Capabilities Development Command Army Research Laboratory (DEVCOM ARL) are pioneering advancements in ground and aerial autonomous systems, as well as energy solutions, to bolster the mobility and maneuverability of these technologies, the technology website reports. 'We are bridging the gap between humans and robots, making them more intuitive, responsive, and, ultimately, more useful for the Soldier,' said a lead researcher for the Artificial Intelligence for Maneuver and Mobility program. 'ARL researchers have demonstrated an interactive bi-directional communication system that enables real-time exchanges between humans and robots.' And of course (CGI): Also read: China And India Drive Global Demand For Air Conditioning
Yahoo
12-07-2025
- Entertainment
- Yahoo
Who Is FOUNDATION's the Mule? The Powerful and Mysterious Baddie, Explained
Season three of Foundation is here, and with it comes the long-awaited arrival of the Mule. The super-baddy has been a distant threat, slowly approaching in the background, much like Thanos over the second half of the Infinity Saga. When he arrives on the scene, the Mule has a similar cataclysmic impact on the story as the Mad Titan of Marvel, too. Everything inevitably bows before his supersized persona and otherworldly power. But who is the Mule in Foundation? Here's what you need to know about him. In the case of the Mule, the fearsome warlord element isn't quite as pronounced. He doesn't use a gigantic sword or don battle armor. Asimov's great villain of the Foundation series wields a different kind of power. It is one that comes not from brawn but from a brain capable of cowing any individual and even entire planets when he wills it. Before we get too far into the details of the man of mystery himself, I want to point out that Apple+ and showrunners David S. Goyer and Bill Bost have spent three seasons adapting what many have called 'unadaptable' source material. And despite a host of necessary shifts in the storylines, they're killing it. Trust me. As someone who has read this series more than once, this isn't an easy task. That said, the Mule is arguably the best part of the story, and the show is actually adapting his part closer to the source material than any other part of the show thus far. What does this have to do with who the Mule is? Let's ask the author himself… Asimov cloaked the character of the Mule in intense mystery in his novels. This means it's hard to know much about him, even as his part in the story expands. For some context, here's how the guy is introduced in the book Foundation and Empire. 'The Mule …. Less is known of 'The Mule' than of any character of comparable significance to Galactic history. Even the period of his greatest renown is known to us chiefly through the eyes of his antagonists and, principally, through those of a young bride… — Encyclopedia Galactica.' The Mule is built to be an unknown. He doesn't get a name, and most people never see him or know what he looks like. Despite the facade, over the course of the original story, we do find out a few key factoids about the Mule. One is that he has fierce telepathic ability (something called a 'mentalic' in Asimov's world). He can mess with people's minds and can suppress the thoughts and hopes of entire planets. This makes it easy to attack his enemies, as he can turn his fiercest adversaries into puppy-like followers in an instant. This is going to be important, as the warlord assaults Foundation and Empire alike in the show, upending Seldon's plans in the process. The Mule's home world is also a key part of his story. He hails from Gaia, an utterly unique planet of mentalics founded and shaped by robot wisdom. (Its origin is altered a bit in the show.) Its people are raised to be telepathic and have learned to connect with the flora and fauna. This turns the planet into a super-organism that plays a critical role in the later stages of the Foundation story. As far as the Mule is concerned, he leaves Gaia when he rejects the collective, turning his back on the 'together we are better than apart' philosophy as groupthink mumbo jumbo and choosing to use his mental powers to cow and conquer others. Again, this is a very important point for later in the story. RELATED ARTICLE Lee Pace Is a Ripped Messiah in FOUNDATION Season 3 Trailer But for now, suffice it to say that the Mule is an exiled mutant who is despised in his homeland. One other thing worth pointing out? He's not who you think he is. In both the show and the books, the Mule is deceptive and clever. Keep an eye on this one (if you can) as Foundation season three plays out, or the Mule pull the wool over your eyes, along with everyone else's.
Time of India
10-07-2025
- Entertainment
- Time of India
Stuck at home this rainy weekend? Latest movies and web series to watch on OTT platforms like Netflix, Prime Video and more
Latest OTT releases this week: Spending this rainy weekend at home? No worries, there's a lineup of fresh content ready to keep you entertained. From brand-new movie premieres to beloved series making their comeback, OTT platforms are brimming with must-watch titles. Among the latest releases are Special Ops season 2, the much-anticipated drama Aap Jaisa Koi, and more. Whether you love thrillers, romances, or feel-good comedies, popular streaming services like Netflix, Prime Video, JioHostar, and others have something for every mood. So settle in with your favorite snacks, get comfortable, and dive into the latest releases waiting to be discovered. Latest movies and web series to watch on OTT platforms Ballard - 9th July 2025, Prime Video Detective Renée Ballard finds herself tangled in a deadly mix of murders and corruption as she tracks a brutal serial killer, only to reveal a dark conspiracy within the police force that could shatter everything she believes in. Foundation Season 3 - 11th July 2025, AppleTV+ While the first two seasons of Foundation didn't strictly follow the original Asimov novels, Season 1 did align in parts with Foundation, and Season 2 introduced characters from the second book, Foundation and Empire. In Season 3, the story delves deeper into the latter half of Foundation and Empire and also begins exploring material from the third novel, Second Foundation. Narivetta - 11th July 2025, AppleTV+ The film offers a worldwide look at state-backed violence carried out by police forces, highlighting how institutions designed to safeguard the public can instead become tools of oppression, especially targeting marginalized groups. Aap Jaisa Koi - 11th July 2025, Netflix Featuring R. Madhavan and Fatima Sana Shaikh, Aap Jaisa Koi is a charming romantic drama that follows the journey of Shrirenu Tripathi, an unassuming middle-aged Sanskrit professor from Jamshedpur. His routine life takes an unexpected turn when he finds himself falling for Madhu Bose, a French teacher. Special Ops: Season 2 - 11th July 2025, JioHotstar Special Ops 2 delves into the escalating dangers of cyber terrorism and digital warfare. Set amidst tense geopolitical conflicts, the series reflects real-world fears over national security at a time when cyberattacks have the power to disrupt and destabilize whole countries. Katie Taylor vs. Amanda Serrano 3 - 12th July 2025, Netflix Following a week filled with open workouts, tense press conference stare-downs, and growing excitement, women's boxing's most intense rivalry comes full circle. On Friday, July 11, Taylor vs. Serrano 3 will be broadcast live worldwide on Netflix from Madison Square Garden in New York.
