AI News Roundup – April 04, 2025
Open Deep Search arrives to challenge Perplexity and ChatGPT Search
Sentient Foundation has developed Open Deep Search (ODS), an open-source framework that equips large language models with reasoning agents to enhance web search capabilities, challenging proprietary AI search tools
China claims AI chip progress as Loongson unveils CPUs amid self-sufficiency push
Chinese semiconductor company Loongson Technology has released new CPUs with its own architecture and integrated GPUs, aimed at enhancing computing performance and supporting China’s goal of IT self-sufficiency
Decentralized platform is letting users own a piece of the AI models trained on their data
Vana is a decentralized platform that allows users to collectively contribute data, govern its use, and own AI models trained on their data, aiming to democratize AI development and benefits
New in NotebookLM: Discover sources from around the web
Google’s NotebookLM now includes “Discover Sources,” a feature powered by Gemini that allows users to find and add relevant web content to their notebooks by describing their topic of interest
China’s love of open-source AI may shut down fast
China’s embrace of open-source AI, driven by US tech restrictions, may be temporary as its AI capabilities grow, potentially leading to a shift away from freely sharing sensitive technology
Mind Brain Lecture: Neuroscientist Anthony Zador on the Future of AI and the Brain
A neuroscientist argues that understanding the brain’s computational principles is crucial for future AI advancements, proposing a NeuroAI approach to overcome AI’s current limitations
Coal-fired power plant, now retired, to become massive gas-powered campus for AI, data centers
A retired coal-fired power plant in Pennsylvania is being transformed into a $10 billion natural gas-powered campus for AI and data centers, driven by the energy demands of Big Tech
It’s time to start preparing for AGI, Google says
Google DeepMind calls for proactive planning regarding AGI, emphasizing the need for AI safety to prevent potential harms and highlighting a growing tension between rapid development and responsible deployment
Yuval Noah Harari: ‘How Do We Share the Planet With This New Superintelligence?’
In an interview, Yuval Noah Harari discusses the potential singularity, the agentic nature of AI, and the challenges of sharing the planet with superintelligence, emphasizing the urgent need for caution and self-correcting mechanisms
OpenAI Plans to Release ‘Open-Weight’ Model with Reasoning Capabilities
OpenAI plans to release a new “open-weight” language model with reasoning abilities, allowing fine-tuning but not access to training data or architecture, marking a shift towards more open access amid increasing AI competition

