Can carnivorous plants eat people? More information Adding us as a Preferred Source in Google by using this link indicates that you would like to see more of our content in Google News results. Venus flytraps are only native to the coastal bogs, wetlands, and pine savannas of North Carolina and South Carolina in the United States. Anyone who's seen the 1986 musical comedy, is familiar with Audrey II . Audrey II looks like a towering, blood-eating Venus flytrap plant, even mimicking the jaw-like, snapping pod of real carnivorous plants. But while Venus flytraps consume insects, Audrey II feeds on humans.

Real-world large language model deployments (e.g., conversational AI systems, code generation assistants) naturally generate abundant implicit user dissatisfaction (DSAT) signals, as users iterate toward better answers through refinements, corrections, and expressed preferences, while explicit satisfaction (SAT) feedback is scarce. Existing preference learning approaches are poorly aligned with this data profile, as they rely on costly human annotations or assume plentiful positive responses. In this paper, we introduce \textbf{DRIFT} (\textbf{D}issatisfaction-\textbf{R}efined \textbf{I}terative pre\textbf{F}erence \textbf{T}raining), which anchors training on real-world DSAT signals and samples positives dynamically from the evolving policy. Empirically, DRIFT models trained on real-world \textit{WildFeedback} datasets and synthetic \textit{UltraFeedback} datasets achieve up to +6.23\% (7B) / +7.61\% (14B) on WildBench Task Score and up to +8.95\% (7B) / +12.29\% (14B) on AlpacaEval2 win rate over base models, outperforming strong baseline methods such as iterative DPO and SPIN. At larger scales, the improvements are particularly pronounced: 14B models trained with DRIFT surpass GPT-4o-mini on WildBench. Further analysis shows that DRIFT also preserves exploratory capacity, yielding more diverse high-reward solutions rather than collapsing to narrow subsets. Theoretically, we demonstrate that this design preserves preference margins and avoids the gradient degeneration. These results show that DRIFT is an effective and scalable recipe for real-world post-training that leverages the most abundant and informative signal. The code and data are available at https://github.com/cacayaya/DRIFT.git.

A nightmarish robot with Venus flytraps for hands that can trap objects in its jaw-like leaves -- and then pick them up -- has been developed by scientists. Engineers from Singapore used tiny remote-controlled electrodes to stimulate severed leaves of the iconic carnivorous plants into closing on command. While the project may seem straight out of the mad scientists' playbook, integrating soft and flexible plant matter into robots could have sensible practical applications. It would allow robots to pick up and manipulate fragile objects that might otherwise be damaged by traditional, mechanical graspers, the team explained. At the same time, it requires less power and responds more rapidly than traditional soft actuators made from polymer-based materials.

There's no denying that human intelligence makes our species stand out from other life on Earth. Our modern brain is an evolutionary feat more than 520 million years in the making, and it is the key to everything that makes us human. But while human brains are extraordinary, we don't have a monopoly on intelligence. "Reserving the term'cognition' for typically human problem-solving abilities ... and dismissing simpler behavior as mechanistic, reflexive, and hard-wired does not do justice to the behavioral complexities of even the simplest of organisms," University of Gronigen psychologist Marc van Duijn and his colleagues write in a widely cited 2006 paper on cognition. You might think of tool use as an exclusively human activity, but macaques on an island off Thailand have learned to use stones as tools to shuck oysters.