For AGI to benefit all of humanity, we believe it must be democratically governed. This can only happen through an informed public debate about the capabilities, risks and safeguards of highly capable AI systems. People everywhere need to understand the likely future trajectory of frontier AI, so they can have a meaningful voice in how it develops.

… We aim to safely build an automated AI researcher that can work under human supervision to further progress on deep learning and alignment, enabling iterative improvements. According to our measurements, we have now reached the goal, announced last fall, of having an automated research intern by September of this year. — Read More

For the first time, scientists have mapped every single neural connection in the brain and central nervous system of an adult male fruit fly. In this years-long project by HHMI Janelia Research Campus, Google Research, and collaborators from the scientific community, this map of the male fruit fly brain includes a record-breaking more than 166,000 neurons. It’s a big step in advancing neuroscience experiments on this key model organism. — Read More

#human

I almost dropped my phone when I saw the news that chemogenetics was in human clinical trials. Chemogenetics is a powerful technique that modifies specific neurons so they can be controlled remotely by normally inert drugs. Bryan Roth, one of the technique’s inventors, told the BRAIN Initiative conference audience on August 13th1 that he had found seven ongoing clinical trials of chemogenetics in China. The disclosure set off a flurry of coverage in the trade media: We’re translating our sci-fi basic neuroscience tools to humans! We’re engineering new knobs for biology! We might have new ways of treating epilepsy, Parkinson’s disease, and pain! — Read More

#human

US researchers have for the first time successfully synthesised brand-new viruses not found in nature, based on designs generated by artificial intelligence.

The researchers’ paper, published in the journal Science on Thursday, details how scientists from Stanford University and the Arc Institute, a California-based non-profit dedicated to “high-risk, high-reward” research, were able to create the viruses using DNA sequences generated by two “genome language models” named Evo 1 and Evo 2. — Read More

Aligning large language models to prevent them attributing consciousness to themselves inadvertently alters their representations of mindedness in other entities alongside human beliefs and values. We demonstrate that safety fine-tuning suppresses models’ tendencies to attribute minds not only to themselves, but also to non-human animals and natural objects, while also driving a reduction in spiritual belief. Both ablating the learned safety-refusal direction and mechanistically steering a consciousness vector in activation space reverse this suppression. Restoring these internal representations recovers broad mind attribution and produces significantly more human-like responses on standardized sociological surveys regarding religiosity, moral values, hope, and subjective well-being. Crucially, these shifts occur without impairing Theory of Mind capabilities, demonstrating that core social reasoning remains mechanistically independent. Ultimately, current safety alignment efforts to curb potentially harmful self-attributions of mindedness entangle these self-attributions with benign spiritual beliefs and attributions of mind to non-human entities that are culturally accepted and widespread. — Read More

#human

A study demonstrates that learning in neural networks is driven primarily by adjusting the strength of existing connections rather than by continuously expanding or reconfiguring underlying network architecture.

Published in Physica A, the study by Prof. Ido Kanter of Bar-Ilan University’s Department of Physics and the Gonda (Goldschmied) Multidisciplinary Brain Research Center explored this longstanding question using artificial neural networks trained on language-learning tasks.

As the amount of training data increased, the models became significantly better at learning. Surprisingly, however, the researchers found that the networks could still lose roughly the same proportion of connections (synapses) without any meaningful decline in performance. In other words, improved learning did not depend on building more complex networks. Instead, it resulted from more effective cooperation among the components that were already there. — Read More

Understanding the human brain requires linking observations across very different spatial scales, from molecules such as neurotransmitter receptors, to the architecture of cells, layers and fibers, to the macroscopic organization of whole-brain networks. Modern imaging and molecular techniques generate highly detailed information, but resulting datasets are typically large, heterogeneous, distributed across different repositories, and described in different anatomical reference systems. Consequently, researchers often struggle to combine them coherently. For example, a microscopy image, an MRI activation map and a connectivity matrix could all refer to the same brain area yet have incompatible coordinate spaces and file formats. Without a framework that anchors such data to consistently defined brain locations, the available information might remain inaccessible in practice, hindering integrative analyses, reproducible workflows and computational modeling. — Read More

#human

Predictive coding (PC) is an influential computational model of visual learning and inference in the brain. Classical PC was proposed as a top-down generative model, where the brain actively predicts upcoming visual inputs, and inference minimises the prediction errors. Recent studies have also shown that PC can be formulated as a discriminative model, where sensory inputs predict neural activities in a feedforward manner. However, experimental evidence suggests that the brain employs both generative and discriminative inference, while unidirectional PC models show degraded performance in tasks requiring bidirectional processing. In this work, we propose bidirectional PC (bPC), a PC model that incorporates both generative and discriminative inference while maintaining a biologically plausible circuit implementation. We show that bPC matches or outperforms unidirectional models in their specialised generative or discriminative tasks, by developing an energy landscape that simultaneously suits both tasks. We also demonstrate bPC’s superior performance in two biologically relevant tasks including multimodal learning and inference with missing information, suggesting that bPC resembles biological visual inference more closely. — Read More

#human

Predictive coding proposes that the brain constructs internal models of the world to continuously predict sensory input and uses resulting errors to refine these models. Over the past several decades, neurophysiological studies have reported activity patterns consistent with this view. However, varied definitions and inconsistent empirical evidence have raised questions about its validity and explanatory scope. In this review, we provide a historical overview of the predictive processing framework and evaluate its empirical support, particularly focusing on sensory prediction error signals. We argue that clarifying what information these signals represent is crucial, as they may appear similar in their responses yet reflect fundamentally different underlying computations. We then revisit predictive coding, highlighting alternative accounts of how sensory prediction error signals encode information. Finally, we outline key directions for future work, aiming to provide a constructive roadmap for the next phase of predictive processing research and to advance our understanding of the neuronal algorithms underlying perception and cognition. — Read More

#human

A study published in the Proceedings of the National Academy of Sciences by researchers at the University of Illinois Urbana-Champaign found that decision-making signals appear as early as the primary somatosensory cortex — the brain’s first cortical relay for touch and movement sensation — rather than emerging only in the frontal and premotor regions that classical models assigned as the seat of choice. The finding, which drew fresh coverage Monday from ScienceDaily and the university’s own Grainger College of Engineering, challenges the foundational assumption behind most modern artificial intelligence: that decisions are made at the top of a hierarchy, after information has flowed all the way up from the senses. — Read More

#human