Latest news with #MicrosoftResearch
The Hindu
4 days ago
- Business
- The Hindu
Watch: AI for All: Dream of democratised and ethical AI
At The Hindu Huddle 2025, experts on the panel 'AI for All' said AI isn't here to replace humans but amplify our capabilities. Kalika Bali of Microsoft Research reminded us that tech mirrors us — both our flaws and brilliance. Astha Kapoor flagged BPO jobs as vulnerable but pointed to opportunities in data management and regulation. Tanvi Lall said AI can liberate human attention for creativity and critical thinking — if we know how to wield it. Read more: 'AI will amplify, not replace, human capacity'
Indian Express
5 days ago
- Science
- Indian Express
Microsoft's AI model Aurora can now predict air quality at high speed and precision
Microsoft's foundational AI model for weather forecasting has been upgraded and can now accurately predict air quality as well. Aurora has been developed by Microsoft Research to forecast a range of weather-related phenomena such as hurricanes, typhoons, etc, with greater precision and speed than traditional meteorological methods, the company said in a blog post earlier this week. It has also published a research paper on Aurora in science journal Nature. Microsoft further said that Aurora's source code and model weights are now publicly available. A specialised version of the model that produces hourly forecasts, including for clouds, has been integrated into the MSN Weather app. The Windows maker has claimed that Aurora is one of the top-performing AI models in the field of weather forecasting. 'What sets Aurora apart is that it is originally trained as a foundation model and can then be specialized through finetuning to go beyond what is considered traditional weather forecasting, such as air pollution prediction,' Microsoft said. 'Because the model first learns from a large and diverse set of data, it can be fine-tuned with smaller amounts of air quality data,' it added. Aurora has been trained on over a million hours of data captured by satellites, radar and weather stations as well as past weather simulations and forecasts, the company said. The AI model can be fine-tuned using additional data to provide forecasts about specific weather events. Its underlying encoder architecture helps to translate massive amounts of data drawn from multiple sources into a standard format that the AI model uses to make predictions. 'We're not putting in strict rules about how we think variables should interact with each other. We're just giving a large deep-learning model the option to learn whatever is most useful. This is the power of deep learning in these kind of simulation problems,' Megan Stanley, a senior researcher with Microsoft Research, said. Microsoft claimed that its Aurora AI model accurately predicted the landfall of Typhoon Doksuri in Philippines four days in advance and better than some expert predictions. The model also successfully predicted a sandstorm in Iraq two years ago. It beat the US National Hurricane Center by providing accurate five-day forecasts of tropical cyclone paths in 2022 and 2023, as per the company. Aurora, which draws compute power from graphics processing units (GPUs), provides weather forecasts in seconds compared to hourly predictions by traditional weather systems running on supercomputers. While the initial cost involved in training Aurora was high, Microsoft said its operational expenses are lower than traditional weather forecast systems. AI weather models like Aurora are not entirely new. Over the past few years, Google DeepMind has released several AI models designed for weather forecasting such as WeatherNext.
Express Tribune
6 days ago
- Science
- Express Tribune
Microsoft says new Aurora AI model can accurately forecast weather changes
Microsoft says new Aurora AI model can accurately forecast weather changes Microsoft says its latest AI system, Aurora, can predict typhoons, hurricanes, sandstorms, and air quality events more accurately and significantly faster than traditional forecasting methods, according to a new research paper and blog post released this week. The company published details of the Aurora model in Nature, citing major improvements in forecast accuracy and efficiency. Trained on more than a million hours of satellite, radar, and ground station data, Aurora is designed to anticipate atmospheric events across the globe and can be fine-tuned for specific weather scenarios. A new paper published in Nature explains how Microsoft's Aurora AI foundation model goes beyond weather forecasting to more accurately predict a range of environmental events, from hurricanes and typhoons to air quality and ocean waves. — Microsoft Research (@MSFTResearch) May 21, 2025 In tests, Microsoft says Aurora predicted the landfall of Typhoon Doksuri in the Philippines four days in advance — outperforming several expert forecasts. The model also outpaced the US National Hurricane Center in tracking tropical cyclones over the 2022–2023 season and correctly anticipated a major sandstorm in Iraq in 2022. While AI-based weather forecasting is not new — Google DeepMind's WeatherNext is another example — Microsoft positions Aurora as among the most advanced models publicly available. The company has also released Aurora's source code and model weights, opening it to researchers and developers. Microsoft reports that, despite its large training set and the infrastructure required to build it, Aurora runs highly efficiently. Forecasts are generated in seconds, compared to hours required by legacy supercomputer-based systems. A specialised version of the model will be integrated into Microsoft's MSN Weather app, providing hourly updates and cloud forecasts. The development is seen as a potential game-changer for meteorologists and climate scientists, offering faster and possibly more accurate tools for understanding extreme weather patterns and air quality.
The Hindu
10-05-2025
- Business
- The Hindu
The Hindu Huddle 2025: 'AI will amplify, not replace, human capacity'
Paranoia around rapid advancements in AI is misplaced and technology would only amplify, and not replace human capacity, experts at a panel discussion said. The discussion titled AI for All: Dream of Democratised and Ethical AI was moderated by author Manu Joseph and had Astha Kapoor, Co-founder, Aapti Institute, Kalika Bali, Senior Principal Researcher, Microsoft Research, India and Tanvi Lall, Strategy, People + AI as experts. To think of AI as an instrument of absolute and malicious power was wrong, said Ms. Bali. 'Technology can amplify human efforts both good and bad. There are questions that we need to manage right now - such as whether the data that is used to train AI is representative of the global majority, how do we ensure that models are aware of the context, how to reduce the risk of biases etc.' Also Read | The Hindu Huddle 2025 Day 2 live updates The nature of jobs will change and new jobs will be created, but the new ones also will be redundant soon as technology advances, pointed out Ms. Kapoor. 'The most vulnerable jobs right now are those in the BPO industry. There will new jobs in managing the data is being generated. There will be chaos and is need for regulation, social security, skill development and preparedness,' she said. Ms. Lall said AI has opened up the opportunity for learning for the curious people. AI and technology can take care of a range of routine work, and humans can turn their attention to creative and critical areas. The experts were of the opinion that in this phase of rapid changes, the human capacity to think critically and make sense of the cognitive overload will be of value. The Hindu Huddle 2025 is presented by Sami-Sabinsa Group Co-powered by: Government of Karnataka, Government of Telangana; Associate Partners: ONGC, Presidency University, TAFE, Akshayakalpa Organic; Energy Partner : Indian Oil Corporation Limited; Realty partner: Casagrand; Knowledge partner: Amrita Vishwa Vidyapeetham; State partner: Meghalaya tourism and Haryana government; Luxury car partner: Toyota; Radio partner: Radio City; Gift partner: Anand Prakash; Broadcast partner: Times Now; Outdoor media partner: Signpost India.
The Hindu
02-05-2025
- Business
- The Hindu
The Hindu Huddle: Can ethics keep pace with the rapid evolution of AI?
In theory, it is already possible for an individual to create facial recognition software using Artificial Intelligence — software that could scrape billions of images from the Internet without permission, recognise faces, and identify individuals. Some of the large AI tools today are capable of doing this, but they refrain for ethical reasons. In that sense, ethics is currently woven into AI, but that may not remain the case as smaller, decentralised AIs emerge to meet growing market demand. AI can be used for surveillance, for generating deepfakes, and for spreading disinformation — all areas already rife with ethical concerns. It can also embed and automate bias, especially in hiring. The unspoken truth is that hiring often wants to be biased. AI can enable this subtly, simply by tweaking a few parameters. There is also an innate human need to always know whether one is interacting with a human or a machine. Yet it is often in a company's interest to conceal that the engagement is with AI. For instance, people may be fine living in homes designed by AI, but they need to know that. They would be far less comfortable reading novels entirely written by AI, even if they are works of pulp fiction. In the near future, publishers may release AI-generated books created in a single day, while actively hiding that fact. This raises the need for an industry standard to govern such practices. But is that even possible? Can ethics keep pace with the rapid evolution of AI? Ethics, after all, is a broad, ambiguous set of rules based on what we know. The future will present wholly unfamiliar scenarios. Who, then, will set the ethical rules for what is yet to come? We like to talk about ethics because it represents a part of morality that remains outside the purview of the State. We want to be seen as good people — while keeping the law and the government out of our businesses. But that is also why ethics lacks teeth. Morality, outside religion and the law, is just statement of intent. Legality, by contrast, is a formalised version of ethics — those ideas a society has agreed upon and encoded into law. But legality implies regulation, and regulation implies government. So, how willing will businesses be to articulate ethical standards, knowing that doing so may invite legal oversight? These are some of the questions that will be discussed during The Hindu Huddle session titled 'AI for all? The dream of a democratised and ethical technology'. The participants are Astha Kapoor, co-founder, Aapti Institute; Kalika Bali, senior principal researcher, Microsoft Research, India; and Tanvi Lall, Strategy, People+AI, and the conversation will be moderated by Manu Joseph, author, columnist, and screenwriter. The Hindu Huddle will be held on May 9 and 10 in Bengaluru. If you have any questions that you would like our moderator to ask the panellists, write to us at huddle@
