Breaking AI & Tech News Analyzed

The Discovery of Tool Use in Bees Bumblebees can use tools to solve a problem, according to experiments that demonstrate their remarkably advanced cognitive abilities. The Experiment and Its Findings The bees were given an adapted version of an experiment that, 100 years ago, first demonstrated chimpanzees could work out how to retrieve an out-of-reach banana by stacking boxes. In the latest research, bees were shown to be able to roll a polystyrene ball to a specific location and climb on to it in order to access an artificial flower on a low ceiling. The Implications of the Study The findings challenge the longstanding assumption that insects operate purely on instinct and mindless trial-and-error learning. “Most people think insects are reflex-based machines,” said Dr Olli Loukola, a behavioural ecologist at the University of Oulu, Finland, and senior author. “That they can’t have any emotional states or feel pain. Some people don’t even realise that they have brains.” The Complexity of the Bees' Problem-Solving Abilities In the most basic version of the test, 75% of the bees were successful in reaching the flower. To test whether the bees were really solving the problem, the scientists put them through increasingly complex versions of the challenge. In the final setup, 23 out of 30 bees were successful in recalling the location of the flower and positioning the ball beneath it. The Future of Insect Cognition Research “We are not claiming that bees think like humans,” said Loukola. “But our findings show that miniature brains can generate flexible solutions to novel problems in ways we are only beginning to understand.” Prof Lars Chittka, a behavioural ecologist at Queen Mary University of London, added: “Bees are a model of how much intelligence you can squeeze into a small nervous system … It’s a good reminder of there being a motivation to pay some respect to these other beings.”

The Lead: Google's AI Shift Toward Autonomous AgentsGoogle has launched Gemini 3.5 Flash, a new AI model representing the company's strategic pivot from conversational AI to autonomous agents capable of independently executing complex tasks. This move signals Google's bet that the future of AI lies in systems that can plan, build, and iterate on real work with minimal human intervention, rather than simply answering questions.The Technical Breakthrough: Gemini 3.5 Flash CapabilitiesGemini 3.5 Flash, introduced at Google's annual I/O developer conference, represents the company's strongest AI model yet for coding and autonomous agents. The model can independently execute coding pipelines, manage research projects, and, in internal tests, build an operating system entirely from scratch. This capability was demonstrated on stage when Google engineer Varun Mohan showed agents spawning off to work on separate components before coming together to build a full operating system inside Antigravity, Google's agentic development platform.Performance Benchmarks: Speed and EfficiencyThe model's performance is remarkable, according to Koray Kavukcuoglu, DeepMind's chief technologist. Flash 3.5 outperforms Google's latest frontier model, 3.1 Pro, on nearly all benchmarks, including coding, agentic tasks, and multimodal reasoning. Most notably, it's four times faster than other frontier models, with an optimized version that's 12 times faster while maintaining the same quality. This speed is crucial for agentic work, where multiple AI agents run simultaneously on long-running tasks.The Industry Shift: From Chatbots to Autonomous AgentsThe release of Gemini 3.5 Flash marks a significant industry shift from AI as a conversational tool to AI as an agentic tool. Google is positioning this as the next wave of AI technology, where systems don't just answer questions but actively plan, build, and iterate on real work. This transition is already showing impact among partners, with banks and fintechs automating multi-week workflows and data science teams finding insights in complex data environments. The model can run autonomously for multiple hours, though it will pause for human input at decision points requiring judgment.Future Outlook: Google's AI Ecosystem ExpansionLooking ahead, Google is developing a complementary model, 3.5 Pro, designed to work in tandem with Flash. According to Tulsee Doshi, Google's senior director and head of product, 3.5 Pro will serve as the orchestrator and planner, leveraging Flash as various sub-agents for tasks requiring brute force tool use. Gemini 3.5 Flash is now the default model in the Gemini app and AI Mode in Search, with agentic capabilities coming to Search and powering Gemini Spark, Google's new personal AI agent designed to run 24/7. As Google expands these autonomous capabilities, the company faces increasing scrutiny regarding safety and ethical considerations, particularly following past incidents with AI systems.

The New Tool for Obesity-Related Disease Risk A new tool developed by UK researchers can help identify individuals most at risk of obesity-related diseases, such as type 2 diabetes, gout, and stroke. This tool uses a type of AI called interpretable machine learning to analyze data from nearly 200,000 participants of the UK Biobank project. How the Tool Works The researchers applied the AI tool to data from participants with a BMI of 27 or greater, identifying 20 health, lifestyle, and demographic features that could predict the 10-year risk of 18 different obesity-related complications. These features include age, sex, total cholesterol, and creatinine levels. The Data Analysis The team tested the validity of the tool, dubbed Obscore, using UK Biobank data and datasets from two independent health studies. The results showed that participants with the same age, sex, and BMI can have very different risks for various obesity-related conditions. The Impact Analysis The tool could help inform strategies for prioritizing who should receive weight-loss interventions, particularly in cases where access to NHS treatments is limited. According to Prof Nick Wareham, the tool is not about extending the use of particular therapies, but rather about developing and validating a score that can help with more rational resource allocation. The Prediction The researchers believe that their tool could be useful for prioritizing individuals who would benefit most from weight-loss medications. However, Naveed Sattar, a professor of cardiometabolic medicine at the University of Glasgow, noted that substantial further development and validation will be required before such an approach can be translated into routine clinical practice.