Martin SCHRIMPF aims to build a digital human brain with modern AI. As the Director of the NeuroAI Lab and an assistant professor at EPFL, his research bridges cognitive science, neuroscience, and AI to understand the brain in computational terms.
A core discovery of his work is that optimizing AI for complex tasks to mirror evolutionary performance pressures leads to brain-like representations in-silico. Leveraging this principle, Martin builds state-of-the-art models of primate vision and the human language system with improved resilience to adversarial perturbations and generalization capabilities. Establishing how we understand and evaluate artificial minds, Martin is the architect of Brain-Score, a global open-source platform used to measure how closely AI networks align with neural and behavioral data. Beyond academia, Martin actively translates NeuroAI into real-world applications, advising emerging NeuroAI startups by drawing on his background at MIT, Harvard, TUM, Oracle, and Salesforce.
His approach to the intersection of humans and machines has been featured in the BBC, Science, and Quanta, and is supported by the Schmidt Foundation's AI2050 initiative among others. Furthermore, his commitment to driving societal impact through technology was recognized with the Google.org Impact Challenge prize.