NethoLabs

AI-Neuroethology, Whole-Brain Emulation and AI Safety

AI-driven neuroethology is offering unprecedented opportunities to advance neuroscience by enabling collection and analysis of data spanning months to years while significantly increasing the life quality of research animals and decreasing the numbers required. We can now vastly improve the power of our studies while gaining new insights into anatomical and functional circuits involved in brain disorders.

Foundation Models (FMs) and Whole-Brain Emulators (WBEs) as breakthrough scientific tools. AI-supported research paradigms also advance the science of dynamical models of the brain such as foundation models and whole-brain emulators. FM and WBE paradigms are theoretically and experimentally sound and provide a novel perspective into neural function in health and disease from the whole-organism perspective.

FMs and WBEs advance AI safety. FMs and WBEs are built on interpretable neurobiological dynamics and behaviours - while still running on known architectures such as transformer-based neural networks (NNs). They thus offer a radically transparent substrate for AI alignment with better-understood anatomy and behaviour, enabling causal insights into goal formation, agency, and failure modes - making them critical tools for determining when advanced AI systems may become unsafe for humanity.