Perplexity has redesigned its iPad app to enhance research capabilities, making it more native to iPadOS with features like split-view and a larger interface, aiming to attract students and professionals and grow its paid subscriber base.
Google's Gemini Deep Research tool can now access users' Gmail, Drive, and Chat data to enhance research responses, raising privacy concerns despite Google's assurances that data used for training AI is not derived from connected apps. The system employs multi-step research plans and is comparable to offerings from OpenAI and others, but users are cautioned to avoid sharing confidential information and not to rely on it for professional advice.
Thomas Wolf, co-founder of Hugging Face, argues that current AI models like ChatGPT are unlikely to produce major scientific breakthroughs, as they tend to predict the most likely next word rather than generate novel ideas, contrasting with the contrarian thinking often required for groundbreaking science. He envisions AI as a tool to assist scientists rather than replace the need for true innovation, highlighting existing applications like DeepMind's AlphaFold and startups aiming to push AI further into scientific discovery.
A US-developed AI tool analyzed over 15,200 open-access journals, flagging more than 1,400 as potentially fraudulent, with over 1,000 confirmed as predatory, to help improve the integrity of academic publishing. The system uses website pattern analysis and is intended as a prescreening aid for human reviewers, not a replacement, and future accessibility to universities and publishers is planned.
Scientists have identified 188 new CRISPR systems from bacteria found in diverse environments such as Antarctic lakes, coal mines, and dog saliva. These newly discovered CRISPR systems have the potential to be repurposed as gene editing therapies, diagnostic sensors, or research tools, driving the race to discover and name the natural diversity of CRISPR systems.
Researchers from the UK have created a publicly available database called the "unknome," which compiles information on thousands of understudied proteins encoded by genes in the human genome. The database assigns a "knownness" score to each protein based on factors such as function, conservation, and subcellular compartmentalization. By prioritizing research on proteins with low knownness scores, the database aims to accelerate the understanding of these genes and their potential roles in critical cellular processes. The authors tested the database by studying genes in humans and flies, revealing that many of these understudied proteins contribute to essential functions. The unknome database provides a platform for identifying and selecting important genes of unknown function, bridging the gap in biological knowledge.