Microsoft has partnered with Swiss start-up inait to develop an innovative artificial intelligence model that simulates mammal brain reasoning. The collaboration aims to enhance AI’s capabilities by allowing it to learn from real-world experiences, as opposed to traditional models which rely on pre-existing data. This partnership builds on two decades of digital neuroscience research and is set to impact various industries, including financial trading and robotics.
The alliance between the tech giant and the Lausanne-based inait seeks to replicate biological intelligence to create a new form of AI. Richard Frey, CEO of inait, explained that the company was founded with the goal of mastering the brain’s proven intelligence to develop a novel and powerful kind of AI. “We are now building products that teach digital brains to address major challenges across industries,” he added.
One of the key features of inait’s technology is its ability to learn from real-world experiences rather than relying solely on existing data correlations. This approach is seen as transformative, with the potential to enhance decision-making in fields ranging from financial algorithms to manufacturing robotics. The technology will be incorporated into Microsoft’s AI offerings, bringing cutting-edge cognitive models to customers in various sectors.
In the financial industry, the partnership will focus on creating advanced trading algorithms, risk management tools, and personalised financial advice. In robotics, the goal is to develop machines that can adapt to more complex, dynamic environments, such as those found in industrial manufacturing.
inait’s work stems from a Swiss government-funded initiative that spent over 20 years using brain research to develop accurate digital replicas of mammal brains. The project, led by Henry Markram, inait’s co-founder, involved creating 18 million lines of computer code for brain simulations, primarily based on the mouse brain. Markram emphasised that this “generic recipe” could eventually be used to replicate other species’ brains, including humans.
One of the advantages of this AI approach is its potential to be less energy-intensive and learn more quickly than existing deep reinforcement models. It is also expected to continue evolving once implemented in real-world applications. Despite the complexity of building a full human brain replica, Markram believes that many business products won’t require such detailed models, and the simulations can still provide significant benefits.
The technology developed through this initiative is being made available to researchers through the Open Brain Institute, a non-profit established by Markram. This provides a unique opportunity for scientists to explore neurological conditions such as autism and better understand brain processes.
The research also contributes to the broader scientific understanding of the brain through projects like the map of the adult fruit fly brain, aimed at creating a comprehensive atlas of connectomes—the pathways for information flow between neurons and synapses. While connectomes provide static maps, dynamic simulations offer a more accurate picture of how brain processes work in real time. According to Anton Arkhipov from the Allen Institute, connectomes are like maps of city roads, while simulations offer a realistic view of how traffic moves through the city under various conditions.
