Google DeepMind builds AI model to predict cyclones; Foxconn sends 97% of India iPhone exports to U.S; AMD unveils AI server
Google DeepMind builds AI model to predict cyclones
Google DeepMind and Google Research have launched a new website called Weather Lab that shares their AI weather models today. The model is able to predict cyclone formations, their route, intensity, size and shape up to 15 days ahead, Google said in a blog. The experimental AI model can also generate 50 different scenarios for the storm.
The research teams have also released a new paper with details of the core along with an archive on Weather Lab of historical cyclone track data, evaluation and backtesting. Google also added that during internal testing they found that the model's predictions were as accurate and often more accurate than current physics-based methods, they have also partnered with the U.S. National Hurricance Center (NHC) to evaluate further how effective the model is. The Weather Lab website also shows a comparison between how AI models perform and how the traditional models perform.
Foxconn sends 97% of India iPhone exports to U.S
Nearly all the iPhones exported by Foxconn from India went to the United States between March and May, customs data showed, far above the 2024 average of 50% and a clear sign of Apple's efforts to bypass high U.S. tariffs imposed on China. The numbers, being reported by Reuters for the first time, show Apple has realigned its India exports to almost exclusively serve the U.S. market, when previously the devices were more widely distributed to countries including the Netherlands, the Czech Republic and Britain.
During March-May, Foxconn exported iPhones worth $3.2 billion from India, with an average 97% shipped to the United States, compared to a 2024 average of 50.3%, according to commercially available customs data seen by Reuters. India iPhone shipments by Foxconn to the United States in May 2025 were worth nearly $1 billion, the second-highest ever after the record $1.3 billion worth of devices shipped in March.
AMD unveils AI server
Advanced Micro Devices CEO Lisa Su on Thursday unveiled a new AI server for 2026 that aims to challenge Nvidia's flagship offerings as OpenAI's CEO said the ChatGPT creator would adopt AMD's latest chips. Su took the stage at a developer conference in California, called 'Advancing AI' to discuss the MI350 series and MI400 series AI chips that she said would compete with Nvidia's Blackwell line of processors.
The MI400 series of chips will be the basis of a new server called 'Helios' that AMD plans to release next year. The move comes as the competition between Nvidia and other AI chip firms has shifted away from selling individual chips to selling servers packed with scores or even hundreds of processors, woven together with networking chips from the same company. The AMD Helios servers will have 72 of AMD's MI400 series chips, making them comparable to Nvidia's current NVL72 servers.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Economic Times
5 minutes ago
- Economic Times
ChatGPT took on a 50-year-old Atari — and lost
Synopsis In a surprising turn of events, ChatGPT, a leading AI chatbot, was defeated by the vintage Atari 2600 in a chess match. Despite ChatGPT's initial confidence and claims of chess prowess, the Atari console, launched in 1977, consistently outperformed the AI. The experiment highlighted the limitations of ChatGPT in logical reasoning and board awareness, leading to its eventual concession.
&w=3840&q=100)
Business Standard
14 minutes ago
- Business Standard
Bhavish Aggarwal's Krutrim bets on India-first AI to rival global peers
Krutrim, the artificial intelligence startup founded by Ola's Bhavish Aggarwal, is positioning its recently launched flagship assistant, Kruti, to stand apart from global peers like OpenAI's ChatGPT and Google's Gemini by leveraging deep local integration, multilingual capabilities, and agentic intelligence tailored to India's unique digital ecosystem. The company calls Kruti India's first agentic AI, capable of booking cabs, paying bills, and generating images while supporting 13 Indian languages using a localised large language model. In the Indian context, the firm competes with global AI giants such as OpenAI, Anthropic and Google, as well as local players such as Sarvam AI and 'Our key differentiator will come with integrating local services,' said Sunit Singh, Senior Vice-President for Product at Krutrim. 'That's not something that will be very easy for global players to do.' Krutrim has already integrated India-specific services, with plans to scale this integration further. The strategy aims to embed Kruti deeply into Indian digital life, allowing it to perform functional tasks through local service connections. This is an area where international competitors may struggle due to regulatory and infrastructural complexities in the Indian market. Voice-first As Krutrim positions Kruti to serve India's linguistically diverse population, the company is doubling down on voice-first, multilingual AI as a core enabler of scale and accessibility. Navendu Agarwal, Group CIO of Ola, emphasised that India's unique language landscape demands a fundamentally different approach from Western AI products. 'India is a voice-first world. So we are building voice-first models,' Agarwal said, outlining Krutrim's strategy to prioritise natural, speech-driven interactions. Currently, Kruti supports voice commands in multiple Indian languages, with plans underway to expand that footprint. Agarwal said the long-term vision is to enable seamless, speech-based interactions that go deeper into local dialects. The company's multilingual, voice-first design is central to its go-to-market strategy, especially in reaching non-English speakers in semi-urban and rural India. The plan also includes integrating with widely used Indian services and government platforms. Krutrim's long-term vision for Kruti centres on true agentic intelligence, where the assistant can act autonomously on behalf of users. Whether it's 'book me a cab to the airport' or 'order my usual lunch', Kruti understands intent and executes tasks without micromanagement. 'Think about it—a super agent which can do food, do apps, provide you help and education information and which can also manage your budget and finance,' said Agarwal. 'So that's what is a mega-agent, or the assistant which is communicating with all of them seamlessly wherever it is needed.' Hybrid technology Rather than relying solely on a single in-house model, Krutrim has opted for a composite approach aimed at optimising accuracy, scalability and user experience, according to Chandra Khatri, the company's Vice-President and Head of AI. 'The goal is to build the best and most accurate experience,' Khatri said. 'If that means we need to leverage, say Claude for coding, which is the best coding model in the world, we'll do that.' Kruti is powered by Krutrim's latest large language model, Krutrim V2, alongside open-source systems. The AI agents evaluate context-specific needs and choose from this suite of models to deliver tailored responses. Investments Krutrim reached unicorn status last year after raising $50 million in equity during its inaugural funding round. The round, which valued the company at $1 billion, included participation from investors such as Matrix Partners India. Earlier this year, company founder Bhavish Aggarwal announced an investment of ₹2,000 crore in Krutrim, with a commitment to invest an additional ₹10,000 crore by next year. The company also launched the Krutrim AI Lab and released some of its work to the open-source community. As Krutrim's AI assistant begins to interface with highly contextual and personal user data, the company emphasises a stringent, India-first approach to data privacy and regulatory compliance. The company employs internal algorithms to manage and isolate user data, ensuring it remains secure and compartmentalised. While Krutrim is open to competing globally, it remains committed to addressing India's market complexities first. 'We don't shy away from going global. But our primary focus is India first,' Agarwal said. Krutrim's emphasis on embedded, action-oriented intelligence—capable of not just understanding queries but also fulfilling them through integrations—could define its edge in the increasingly competitive AI landscape. Here, localisation and service depth may become as critical as raw model power.
NDTV
an hour ago
- NDTV
Nearly 7,000 UK University Students Caught Cheating Using AI: Report
Nearly 7,000 university students in the UK were caught cheating using ChatGPT and other artificial intelligence tools during the 2023-24 academic year, according to data obtained by The Guardian. As part of the investigation, the British newspaper contacted 155 universities under the Freedom of Information Act. Of those, 131 institutions responded. The latest figures show 5.1 confirmed cases of AI-related cheating for every 1,000 students, a rise from 1.6 per 1,000 the previous year. Early projections for the current academic cycle suggest the number could climb even higher to 7.5 per 1,000 students. The growing reliance on AI tools like ChatGPT is proving to be a major challenge for higher education institutions. At the same time, cases of traditional plagiarism have dropped. From 19 per 1,000 students in 2019-20 to 15.2 last year, the number has gone down and is expected to fall further to 8.5 per 1,000. Experts warn that the recorded cases may be only scratching the surface. "I would imagine those caught represent the tip of the iceberg," said Dr Peter Scarfe, associate professor of psychology at the University of Reading. "AI detection is very unlike plagiarism, where you can confirm the copied text. As a result, in a situation where you suspect the use of AI, it is near impossible to prove, regardless of the percentage AI that your AI detector says (if you use one). This is coupled with not wanting to falsely accuse students." Evidence suggests AI misuse is far more widespread than reported. A February survey by the Higher Education Policy Institute found that 88 per cent of students admitted to using AI for assessments. Researchers at the University of Reading tested their own systems last year and found AI-generated submissions went undetected 94 per cent of the time. Online platforms are making it easier. The report found dozens of videos on TikTok promoting AI paraphrasing and essay-writing tools that help students bypass standard university detectors by "humanising" ChatGPT-generated content. Dr Thomas Lancaster, an academic integrity researcher at Imperial College London, said, "When used well and by a student who knows how to edit the output, AI misuse is very hard to prove. My hope is that students are still learning through this process." Science and technology secretary Peter Kyle told The Guardian that AI should be used to "level up" opportunities for dyslexic children. Tech giants are already targeting students as key users. Google offers university students a free 15-month upgrade to its Gemini AI tool, while OpenAI provides discounted access to students in the US and Canada.
