Northwestern engineers printed artificial neurons from molybdenum disulfide and graphene that fire at precisely the same speed as biological neurons, enabling them to reliably activate real brain cells in mouse tissue, something previous artificial neurons failed at. The critical discovery is that firing-rate matching is necessary for biological interfacing, and printable 2D materials now achieve it. Applications span neuroprosthetics, brain-machine interfaces, and energy-efficient neuromorphic computing.
Scientists are building sub-nanometer DNA robots that use strand-displacement chemistry and external fields to navigate inside the body, delivering targeted drugs and capturing viruses like SARS-CoV-2. The machines exploit DNA precision folding to act as programmable molecular nano-surgeons with sub-nanometer positioning accuracy, a feat no classical mechanism achieves. Current challenges involve Brownian motion and isolation in complex biological environments, but the field is advancing rapidly toward fully autonomous in-body drug platforms.
EPFL researchers discovered that noise in quantum circuits erases the contributions of earlier computational steps, causing deep circuits to behave effectively like shallow ones. This means adding more circuit layers does not improve performance because noise wipes out the influence of early operations before they can affect results. The finding reframes the quantum computing roadmap: progress requires noise reduction or inherently noise-resilient circuit design, not simply more circuit depth.
Columbia Engineering Creative Machines Lab built a robot that learned realistic lip movements for speech and singing by first watching its own reflection, then studying hours of YouTube videos with no preset programming required. The system controls 26 facial motors, forms words in multiple languages, and performed a song from an AI-generated debut album. The breakthrough addresses the Uncanny Valley problem, since humans focus nearly half their attention on lip motion during conversation and lifeless robot lips undermine trust.
USC researchers published in Science a memory device using a tungsten/hafnium-oxide/graphene sandwich that reliably retains data at 700C, hotter than molten lava, for over 50 hours without refresh. Graphene repellent surface chemistry with tungsten prevents the electrode-bridging short circuits that destroy conventional devices at high temperatures. Applications span deep-earth drilling, nuclear and fusion systems, space exploration, and next-gen AI accelerators that run far hotter than today chips can tolerate.
Zhejiang University researchers found that Centaur, an AI trained on 10 million human psychological choices and claimed to simulate human cognition, was merely pattern-matching. When given a new instruction to always choose option A, it ignored the directive and continued selecting training-data correct answers, revealing a fundamental gap between benchmark performance and real language comprehension. The finding challenges how AI cognitive simulations are evaluated and exposes overfitting as a masquerade for intelligence.
Princeton researchers grew 70,000 living neurons around a 3D scaffold of microscopic metal electrodes, creating a biocomputer that successfully recognizes both spatial and temporal electrical pulse patterns. Unlike flat petri-dish approaches, this inside-out scaffold interfaces directly with neurons as they develop naturally in 3D, achieving pattern recognition at roughly one-millionth the power of comparable AI systems. The device simultaneously advances study of neurological diseases and charts a path toward bioelectronic AI.
Using OpenEvolve — an open-source LLM-driven evolutionary search tool — researchers at Oratomic found a quantum algorithm requiring only 3 atoms to encode a qubit, a 100x reduction over prior approaches. The AI autonomously bridged niche sub-discipline knowledge across thousands of iterations, acting like accelerated natural selection for algorithm design. The pace of AI-assisted quantum discovery is reportedly outrunning expert prediction.
Researchers created molecular devices that dynamically switch roles — memory, logic gates, or artificial synapses — within the same physical structure depending on chemistry and environment. Unlike conventional solid-state electronics with fixed architectural roles, these molecules can unlearn, pointing toward hardware that reconfigures like a biological brain. The researchers described the versatility as a genuine surprise.
A hybrid neuro-symbolic system combining neural networks with formal symbolic reasoning achieved up to 100x energy reduction while improving accuracy on robotic tasks. By encoding logical rules that constrain the neural search space, the system avoids brute-force trial-and-error. Researchers frame this as a fundamental shift away from scaling compute toward scaling reasoning.
Engineers built a nanoelectronic device from modified hafnium oxide that simultaneously processes and stores information in the same physical location — breaking the von Neumann bottleneck that separates memory and compute in every modern CPU/GPU. The device cuts energy use by up to 70% and learns in real time. Intel's Hala Point system at Sandia Labs has already scaled this to 1.15 billion such neurons.
Researchers are building Fungal Computer Interfaces (FCI) creating two-way communication between fungal mycelium and digital systems, exploiting the fact that mycelium transmits electrical spikes structurally similar to neural action potentials. New microelectrode arrays can now record from hundreds of network points simultaneously for ML analysis. A 2026 study introduced MycGNN — a graph neural network inspired by mycelium's exploration strategies.