UCLA is pioneering the use of ChatGPT Enterprise in higher education. They're inviting proposals for projects using the platform to enhance university operations. Successful projects will receive access to OpenAI accounts. This follows UCLA's agreement with OpenAI, making them California's first university to integrate these capabilities.
Nvidia has launched AI Blueprint for video search and summarisation, combining computer vision and generative AI. Part of Nvidia Metropolis, it allows developers to create AI agents that analyse visual data, answer questions, and generate alerts. Major companies are adopting it for various industries. The technology uses vision language models, can be deployed on Nvidia GPUs in different environments, and is available for free testing with production deployment through Nvidia AI Enterprise.
The UN is launching a global initiative to guide AI use in disaster management, building on existing work. AI is enhancing disaster response through various applications, including improved weather forecasting and resource allocation. Private companies are developing faster, more efficient AI climate models. Challenges include data quality, AI complexity, and cross-border implementation. The initiative will test recommendations with a Mediterranean and European early warning system, while companies address data gaps in underserved regions.
Camelot Secure developed an AI tool called Myrddin to assist with CMMC compliance. Integrated into their dashboard, it provides real-time guidance for IT teams conducting assessments. Using advanced AI, Myrddin aims to simplify and accelerate the CMMC process, reducing time and errors. Camelot received an award for this innovation and plans to expand the tool's capabilities in the future.
Six security vulnerabilities were found in Ollama, an open-source AI framework for local deployment of large language models. These could enable attacks including denial-of-service, model poisoning, and theft. Four have been fixed in recent versions, but two remain unpatched. With nearly 10,000 internet-facing instances, about 25% are vulnerable. Maintainers suggest using a proxy or firewall to filter exposed endpoints.
Companies across sectors are heavily investing in conversational AI powered by large language models, with some firms projected to spend tens of millions annually. A comprehensive list of 50 companies' AI usage, models, and expenditure has been compiled.
A New York AI company has secured £4 million in Series A funding to develop loan origination software for mid-tier US banks and credit unions. This investment is part of a growing market projected to reach £12.2 billion by 2032.
Notion, an AI productivity startup, has leased a large office space in downtown San Francisco for the next decade, with room to expand. This move is part of a trend of AI firms taking up office space in the city, despite high vacancy rates. Other AI companies like OpenAI and Anthropic have made similar moves recently.
Meta has developed Cicero, an AI system that excels at the board game Diplomacy. Cicero combines language models and strategic reasoning to communicate naturally with humans while making effective decisions. It achieved top 10% performance against human players online. This advance in strategic AI demonstrates progress towards handling realistic scenarios involving human interaction, though challenges remain for general-purpose applications.
SK Hynix is fast-tracking development of its next-gen AI memory chips (HBM4), aiming to ship six months earlier than planned in late 2025. This acceleration responds to high demand from Nvidia, which still can't meet current AI chip demand.
Samsung's new Galaxy AI suite includes Sketch to Image, a tool that transforms rough doodles into polished images. Available on new Galaxy Tab models, it has safeguards against inappropriate content generation and can add details not present in the original sketch. The tool refused to generate images from suggestive sketches until certain elements were removed.
Meta FAIR and university researchers developed Thought Preference Optimization (TPO), a method to improve large language model responses. TPO enhances the model's internal thought process before answering, leading to more accurate and coherent outputs. It outperforms baseline models on benchmarks and shows potential for various applications, including creative tasks.
An $11 billion debt market has emerged for AI companies buying Nvidia chips, raising concerns about Nvidia's AI market dominance. This funding targets 'neocloud' groups, likely new AI-focused cloud computing companies, reflecting Wall Street's enthusiasm for AI technologies.
Generative AI art is evolving, but humans remain essential for creation and interpretation. This technology is changing how people interact with and create art, raising questions about human creativity and agency in AI-generated artwork.
A study predicts AI-related e-waste could grow from 2,600 tonnes in 2023 to 2.5 million tonnes by 2030, potentially reaching 1.2 to 5 million tonnes cumulatively between 2020 and 2030. North America is expected to produce the majority, followed by East Asia and Europe. US restrictions on GPU sales to China may exacerbate the issue. Researchers propose a circular economy strategy to potentially reduce AI-related e-waste by 86% globally, including extending hardware lifespan, reusing components, and improving computing efficiency.
A Chinese autonomous driving company has raised $100 million and aims to have its system in 200,000 cars by 2025, up from 20,000 now. The system is comparable to Tesla's and could enable affordable smart EVs in China. The company's mapless approach gives it a cost advantage.
Meta Platforms' AI strategy has gained support from a previously sceptical fund manager. The company plans to heavily invest in AI, potentially improving advertising and messaging capabilities. This strategy has received mixed reactions from analysts, with some praising its potential impact on advertising performance, while others remain cautious about its monetisation.
AI systems, including deep learning and neural networks, are based on mathematics. Neural networks process inputs through layers of artificial neurons, with training involving adjusting connections to minimise errors. Researchers are working to understand neural network decision-making and correct errors. Applications include autonomous satellites, spacecraft inspection, and non-invasive diagnosis of cardiovascular disease.
ChatGPT can assist with holiday travel savings in four ways: 1) Estimating trip costs to identify areas for reduction. 2) Comparing accommodations and recommending affordable options. 3) Suggesting efficient packing to avoid baggage fees. 4) Optimising travel rewards and credit card points for potential savings.
