122
Google Develops AI Co-Scientist to Aid Researchers
Date: February 19, 2025Source: The IndependentGoogle has revealed a new system called 'Co-Scientist' that it says could lead to major new breakthroughs in scientific research. The AI tool is powered by similar technologies to those in chat-based large language models and is specifically devoted to helping find new research breakthroughs. Scientists who have used it describe it as being like having an especially well-read and helpful colleague, capable of synthesizing existing knowledge and proposing new hypotheses and possible experiments.Link: https://www.the-independent.com/tech/google-ai-artificial-intelligence-science-b2700881.htmlMicrosoft develops AI model for videogames
Date: February 20, 2025Source: ΣΕΠΕMicrosoft has developed an artificial intelligence model for videogames, which can help generate visuals and actions, the company announced on Wednesday.Link: https://www.sepe.gr/en/business/22540700/microsoft-develops-ai-model-for-videogames/iPhone 16e: Apple gambles on AI features at lower cost
Date: February 19, 2025Source: eweek.comApple has launched the iPhone 16e, a budget-friendly iPhone that includes Apple Intelligence, a generative AI suite. The device starts at $599 and features a 6.1-inch display, 48MP camera, and Apple's C1 5G modem. It supports AI features like natural language queries and quick photo editing, and includes a customizable Action button and FaceID. The iPhone 16e is designed to bring AI capabilities to a wider audience at a lower cost.Link: https://www.eweek.com/news/apple-iphone-16e-ai/EU accused of leaving ‘devastating’ copyright loophole in AI Act
Date: February 19, 2025Source: The GuardianThe EU is facing criticism for a 'devastating' copyright loophole in its AI Act. Critics argue that the law fails to safeguard the rights of European creatives, leaving them vulnerable to exploitation by tech companies. The TDM exemption has been particularly contentious, with many fearing that it allows large companies to freely harvest intellectual property without fair compensation or proper recognition for the original creators.Link: https://www.theguardian.com/technology/2025/feb/19/eu-accused-of-leaving-devastating-copyright-loophole-in-ai-actOpenAI weighs special voting rights to guard against hostile takeovers, FT reports
Date: February 18, 2025Source: Business InsiderOpenAI is considering granting special voting rights to its non-profit board to preserve the power of its directors, as the ChatGPT-maker fends off an unsolicited takeover bid from Elon Musk. The move could help OpenAI block future hostile takeover attempts, including from Tesla CEO Musk, who co-founded the company with CEO Sam Altman but later departed.Link: https://markets.businessinsider.com/news/stocks/openai-weighs-special-voting-rights-to-guard-against-hostile-takeovers-ft-says-1034372668BBC News finds that AI tools ‘distort’ its journalism into a confused cocktail with many errors
Date: February 14, 2025Source: BBC NewsThe BBC tested four generative AI tools on its own site and found many 'significant issues' and factual errors. The AI assistants, including OpenAI's ChatGPT, Microsoft's Copilot, Google's Gemini, and Perplexity, were given access to the BBC website and asked questions about the news. The AI answers were reviewed by BBC journalists and found to contain significant inaccuracies and distorted content. Over half of the AI answers had significant issues, 19% introduced factual errors, and 13% of quotes sourced from BBC articles were altered or not present in the original source.Link: https://www.niemanlab.org/2025/02/bbc-news-finds-that-ai-tools-distort-its-journalism-into-a-confused-cocktail-with-many-errors/What is sparsity? DeepSeek AI’s secret, revealed by Apple researchers
Date: February 19, 2025Source: ZDNetThe AI market was rocked by DeepSeek, an open-source large language model developed by a China-based hedge fund. DeepSeek's success is due to its use of sparsity, a technique that eliminates parts of the data not affecting the model's output. Apple AI researchers explained how DeepSeek and similar approaches use sparsity to achieve better results with less computing power. Sparsity involves turning off parts of the neural network, reducing parameters, and improving efficiency.Link: https://www.zdnet.com/article/what-is-sparsity-deepseek-ais-secret-revealed-by-apple-researchers/Will AI destroy human creativity? No – and here’s why
Date: February 19, 2025Source: ZDNetThe article discusses whether AI can destroy human creativity, concluding that it does not. It highlights the evolution of AI and its ability to amplify human creativity, but notes that true innovation stems from human experience and originality.Link: https://www.zdnet.com/article/will-ai-destroy-human-creativity-no-and-heres-why/US AI Safety Institute will be ‘gutted,’ Axios reports
Date: February 19, 2025Source: ZDNetThe US AI Safety Institute (AISI) is facing significant cuts, with 497 roles expected to be eliminated. This includes 74 postdocs and 57% of CHIPS staff focused on incentives. The cuts will impact AI safety and regulation efforts, as well as semiconductor production initiatives. The AISI was created to oversee AI model testing and partner with developers on safety measures. The director of AISI recently stepped down, and the institute's staff were left out of the AI summit in Paris, indicating a shift in the Trump administration's priorities towards AI dominance over safety and regulation.Link: https://www.zdnet.com/article/us-ai-safety-institute-will-be-gutted-axios-reports/The head of US AI safety has stepped down. What now?
Date: February 20, 2025Source: ZDNetThe head of US AI safety has stepped down, leaving the field of artificial intelligence without a clear leader. This move raises concerns about the future of AI regulation and safety in the United States.Link: https://www.zdnet.com/article/the-head-of-us-ai-safety-has-stepped-down-what-now/This Week in AI: Maybe we should ignore AI benchmarks for now
Date: February 13, 2025Source: TechCrunchTechCrunch discusses the rapid development of AI models, suggesting that benchmarks may not accurately reflect real-world applications. The article mentions Anthropic's upcoming AI model, which is expected to outperform OpenAI's o3-mini-high on some programming tasks.Link: https://techcrunch.com/2025/02/13/anthropics-next-major-ai-model-could-arrive-within-weeks/Sanas taps AI to change call center workers’ accents in real time
Date: February 19, 2025Source: TechCrunchSanas, a company founded in 2020, uses AI to adjust call center workers' accents in real-time, aiming to reduce accent bias and enhance human communication. The technology analyzes speech and outputs converted speech that matches a specified accent, preserving the original speaker's emotion and identity. Sanas has around 50 customers across various industries and has raised over $100 million in capital, with a recent $48 million funding round valuing the company at over $500 million.Link: https://techcrunch.com/2025/02/19/sanas-taps-ai-to-change-call-center-workers-accents-in-real-time/How we test AI at ZDNET in 2025
Date: February 19, 2025Source: ZDNetZDNET tests AI products by evaluating their performance, helpfulness, and privacy. Experts conduct hands-on testing, assessing speed, quality, and everyday use cases. The focus is on in-depth evaluations of large language models, chatbots, and image generators, considering factors like cost and safeguards against deepfakes and copyright infringements.Link: https://www.zdnet.com/article/how-we-test-ai-at-zdnet-in-2025/