 |
|
|
|||||
A well-funded AI lab with a deep bench of research talent is releasing a powerful new model that generates high-definition video for the film and advertising industries. The company, Moonvalley, on Wednesday launched its first model, named Mareya nod to early cinema pioneer Étienne-Jules Mareywhich could soon help Hollywood studios dramatically speed up production and cut costs. What sets Marey apartand has caught the attention of risk-averse studiosis its training data. The model was trained exclusively on video content either owned or fairly licensed by Moonvalley, avoiding the copyright gray zones that make much generative content legally fraught. Moonvalley is rolling out Marey to a group of 20 trusted filmmakers, some of whom work for major, household-name studios (whose names havent been disclosed), according to cofounder and CEO Naeem Talukdar. The version being released is still early-stage, with regular updates and new features planned. The key differentiator, Talukdar says, is Mareys native high-definition outputsomething thats been notoriously difficult to achieve. The challenge is that if you want to output in high definition, your inputs have to be in high definition, so you need to be able to train the model on HD footage, he says. In other words: It demands lots of powerful servers. By contrast, Talukdar says, most other video generation models are trained on lower-quality 480p or 720p videoand even then, they often compress the data before encoding. That compromises the models understanding of fine-grained detail, leading to uncanny or distorted outputs (like misrendered fingers). Marey is designed to overcome that. Its also operated differently. Unlike most consumer-facing video generators that start with text prompts, Marey is tailored for professional workflows. Filmmakers can input storyboards or keyframes; actors can film demo reels on their phones, which producers can then use to generate scenes showing the actor from different angles or performing new movements. The model can make subtle edits, like changing the direction of wind in a characters hair, or adjusting production-quality scenesfilling in background details or background characters (commonly played by extras). Moonvalley is also running pilot programs with brands looking to train the model on their own imagery and style guidelines. That could allow companies to generate broadcast-quality commercials on demand. We have a number of smaller boutique brands who suddenly are like, Hey, I can go and make a Super Bowl commercial, Talukdar says. The startup raised a $70 million seed round last year, backed by Bessemer Venture Partners, Khosla Ventures, and General Catalyst. Vinod Khosla is described as an unofficial member of the board. Beyond the top-tier investors, Moonvalley has serious R&D chops. Cofounders Mateusz Malinowski and Mikołaj Bińkowski previously led video research at DeepMind and helped develop the model that later became Googles Veo 2 video generator. The team includes six other DeepMind alumni, alongside talent from Meta, Microsoft, Google, TikTok, and Snap. While Talukdar acknowledges that AI tools like Marey could lower production costs, he insists its certainly not the beginning of the end for actorsor filmmakers. Budgets, he argues, arent likely to shrink; instead, creators will use the savings to pursue bigger, more ambitious ideas. I think youre going to see new jobs, he says. And more importantly, I think what this is really going to do is empower the creators more than anybody else.
Category:
E-Commerce
For several years now, pharmacies have faced economic challenges, from falling foot traffic to increased online competition. This has led many major pharmacy chains, including Walgreens, CVS, and Rite Aid, to shutter locations and reduce their footprint. As a matter of fact, over the past decade, the United States has lost nearly 30% of its drugstores. But falling foot traffic and increased online competition aren’t the only things leading to a swath of drug store closures across America. New research from the nonprofit American Economic Liberties Project (AELP) shows that in the past three months alone, the United States has seen another 326 pharmacies close. And the report argues that the reason for those closures comes down to Congresss abandonment of pharmacy benefit manager (PBM) reform. What is a pharmacy benefit manager? The term pharmacy benefit manager, better known by its acronym PBM, was little known outside of industry circles until last year, despite PBMs being a major part of Americas privatized healthcare system. A PBM is an entity that acts as a middleman between insurance companies and drugmakers. There are three major PBMs in America, run by UnitedHealth Group, Cigna Group, and CVS Health Corporation. PBMs help negotiate drug rebates from manufacturers, which PBM companies argue helps lower the cost of drugs for consumers. But this is often disputed. As Fast Company previously reported, the Center for American Progress (CAP) has noted that because PBMs may retain a portion of the rebate they negotiate and also may receive more money from an insurer than the drug is worth, the involvement of PBMs in the American healthcare system may be actually leading to higher drug costs overall. Fast Company reached out to all three major PBMs for comment on AELP’s report. We’ll update this post if we hear back. Report says PBMs are harming pharmacies PBMs also have a negative impact on pharmacies themselves, according to AELP. The report says that since the three major PBMs control 80% of U.S. prescription drug claims, they are able to demand untenably lower reimbursement rates from independent pharmacies in exchange for inclusion in their networks. AELP says many pharmacies choose to accept the lower reimbursement rates for fear of losing access to a large share of their customer base. The untenable rates, the nonprofit continues, are accelerating the pharmacy closure epidemic. It supports its arguments by providing a list of 326 pharmacies across America that have closed since December 19, when lawmakers effectively abandoned plans for PBM reform. Between December 19, 2024, and February 28, 2025, at least 237 independent pharmacies and 89 chain pharmacies have shut their doors. And since January 1, 2024, when AELP began tracking closures, at least 3,179 pharmacies have closed for good. AELP says that the closures, which it attributes to PBM practices, dont just impact the pharmacy store owners and workers but the local community as a whole. As predicted, without Congressional intervention, the Big Three PBMs have continued to abuse their market power, squeezing at least 326 pharmacies237 of them independentout of business in fewer than 10 weeks and stranding their most vulnerable patients in pharmacy deserts without access to lifesaving care, said Emma Freer, AELP’s senior policy analyst for healthcare. PBM reform abandoned The December 19, 2024, date that AELP is using as a starting point for tracking the latest pharmacy closures is a significant one. In 2024, PBMs started coming under increased scrutiny as drug costs in America continued to rise. This scrutiny had bipartisan support, with both Democratic and Republican lawmakers introducing legislation to rein in the unchecked operations of PBMs. But then, any bipartisan legislation that was being worked on fell through when, on December 19, Congress abandoned bipartisan, bicameral PBM reforms as part of a stopgap spending bill, AELP notes. In the lead-up to December 19, Elon Musk suddenly began tweeting his opposition to PBM reform. As AELPs Freer notes, Despite later admitting that he does not know what a pharmacy benefit manager (PBM) is, Elon Musk successfully tanked PBM reforms with nearly unanimous House support late last year. The American Economic Liberties Project is releasing its report about the pharmacy closures in the wake of the abandonment of PBM reform to highlight just how needed such reform is for the survival of independent pharmacies and the ability for Americans to access the drugs they need at a price that is both fair and affordable. As Freer states, it is critical that Congress stand up to these healthcare monopolist middlemen and pass structural PBM reforms that will save their constituents time, money, and lives.
Category:
E-Commerce
Satellite-based disaster monitoring has been a slow and tedious process for decades. The process consists of capturing images, transmitting them back to Earth, and relying on human analysts to interpret the data. This often led to first responders receiving critical information, often too late to act effectively. But AI is now revolutionizing satellite operations in space, aiding real-time image processing and autonomous decision-making. NASAs latest space venture focuses on AI-powered autonomous satellites that can operate without human oversight. In collaboration with Ireland-based satellite intelligence startup Ubotica, NASAs Jet Propulsion Laboratory (JPL) has developed Dynamic Targeting, an AI-driven system that allows satellites to process image data onboard, potentially enhancing disaster response. Dynamic Targeting was recently tested in real-world scenarios, including the Palisades Fire in Los Angeles and the Valencia floods. The AI system was integrated in CogniSAT-6 satellite, a 6-unit cubesat developed by Ubotica and NASA JPL, and autonomously processed data onboard and transmitted insights to Earth within minutes. At the core of CogniSAT-6 lies Live Earth Intelligence (LEI)Uboticas onboard processing platform that integrates AI agents into Earth observation satellites. Paired with SPACE:AI, an end-to-end vision processing framework, the architecture transforms conventional satellites into always-connected observers, enabling rapid decision-making. With LEI, we can quickly deploy and run AI models from third parties directly in orbit. By leveraging Inter-Satellite Links (ISL), these insights reach the right people faster, ensuring critical information gets where its needed, exactly when its needed, Fintan Buckley, CEO of Ubotica, told Fast Company. Were at an inflection point. Satellites will no longer just observe; theyll analyse, interpret, and respond in real time. Dynamic Targeting: The Future of AI-Powered Earth Observation CogniSAT-6s Dynamic Targeting system can analyze lookahead images in as little as 50 seconds. Buckley added that if one satellite cant capture the image due to cloud cover, the AI system alerts others to try on their next pass, removing the need for operators to retask satellites manually. In mixed constellations, the system can even switch to radar imaging (SAR) when clouds are an issue, ensuring data is still collected. “With Dynamic Targeting, the satellite first takes a quick, low-resolution ‘look-ahead’ image, and onboard AI analyzes it for cloud cover. If the target area is clear, the satellite locks on and captures a high-resolution image. If not, it discards the request, conserving bandwidth and storage, explained Buckley. During the Palisades Fire in Los Angeles, identifying smoke plumes through autonomous imaging allowed emergency teams to monitor the fires spread effectively. Likewise, during the Valencia floods, the AI swiftly estimated that 21% of the observed area near Valencia was flooded, and sent down accurate flood data immediately to Earth. “AI-powered satellites can analyze a scene and deliver insights to the ground within minutes, making Earth observation viable in situations where other imaging methods fall short,” said Buckley. “These satellites will soon be integrated with insights from other sources to create an accurate, up to date, view of the situation on the ground to support the responders to manage the situation. The partnership between Ubotica and NASAs Jet Propulsion Laboratory (JPL) began in 2022, when they collaborated to test AI-driven image processing aboard the International Space Station (ISS). Under a $632,000 contract with NASAs Jet Propulsion Lab in California, Ubotica is currently preparing for the first live in-orbit test of Dynamic Targeting in early 2025 through CogniSAT-6. U.S. vs. China: The Geopolitical Race for Space Intelligence While NASA and Ubotica are pushing the boundaries of satellite technology, they face competition from China. The country has been aggressively deploying its own AI-powered Earth observation satellites, Tiantuo and Zhuhai. Operated by the China Aerospace Science and Technology Corporation (CASC) and commercial partner Zhuhai Orbita, has already incorporated AI-driven image processing similar to NASA JPLs initiatives. The geopolitical implications of this race are profound, as autonomous satellites can also provide strategic intelligence on environmental changes, maritime security, and even military movements. NASA JPL and Ubotica, however, are actively collaborating with defense agencies across the U.S. and Europe, to enhance Maritime Domain Awareness. Their AI-powered satellites are playing a crucial role in securing maritime assets such as underwater cables, offshore wind farms, and detecting suspicious vessel activity. It is important to safeguard the vast network of underwater high-speed communication cables, as they often subject to accidental or deliberate damage, Buckley added. The key is to identify and warn off vessels before any damage occurs, and if an incident happens, track and hold the offending vessel accountable. However, this leap in technology also raises critical questions about the reliability of AI decision-making in life-or-death situations. Can AI be Trusted Without Human Oversight? Traditionally, Earth observation has relied on human judgment to verify and interpret satellite data. AI-powered autonomous systems could misclassify minor weather shifts as emergencies or overlook critical events due to biased training data. Despite these concerns, Buckley claims that AI autonomy is inevitable. Human oversight will eventually become obsolete, Buckley told Fast Company. But like almost every other disruptive technology, it will take much longer than anticipated for this to happen. To mitigate AI errors, Uboticas Dynamic Targeting system integrates multiple safeguards through its Live Earth Intelligence (LEI) framework. A built-in Neural Network Supervisor constantly monitors AI outputs, discarding insights that fall outside trained parameters. The system also cross-validates insights by incorporating data from multiple sources rather than relying solely on satellite imagery, he explained. The system ensures continuous learning and improvement by actively selecting images to enhance future model performance. NASAs bet on AI is a bold leap into uncharted territory that could redefine how we monitor our planet and respond to crises. As Buckley explained, AIs role isnt just about analyzing satellite imagery; its about coordinating real-time responses. With Dynamic Targeting, we can command other satellites in a constellation to provide real-time updates as a fire develops. Could this capability extend to drones responding to wildfires? Absolutely.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||