Over the last couple of decades, the relationship between neuroscience and artificial intelligence (AI) has borne promising approaches to increasing human’s abilities paving the way for a delightful future of human-machine collaboration.
A recent research paper goes deep into neural representations utilized for human planning. Scientists were interested in the hippocampus, the part of the brain associated with memory and navigation. They concluded that the hippocampus is important in the ‘rehearsal’ of previous events that inform future actions. This process consists of multiple neural activities that are interlinked due to the nature of imagery within the brain where previous information is used to simulate future scenarios.
By developing computational models that contain this feature, they have successfully attempted to mimic how the human brain thinks and plans. The models rely on recurrent neural networks, which are commonly used for sequential data processing, as the imported information has to be structured in terms of rational thought. Researchers have been puzzled for some time about which are the types of decisions that require us to think and plan before acting and how the cognitive load parameter is involved.
The SYMBIOTIK Project: Towards a New Way of Human and AI Interaction
Simultaneously with these discoveries made in neuroscience, the SYMBIOTIK project is working on enhancing the interaction of humans and AI. This project targets creating Symbiotic AI systems – that augment human cognitive skills instead of replacing them. The SYMBIOTIK project deviates from conventional thinking, as its aim is not to outperform humans in certain cognitive functions but to assist them in making better decisions.
In the SYMBIOTIK project, it is clear that for AI to truly augment human abilities, it must operate in harmony with human thought processes. This requires an understanding of how humans strategize, make decisions, and undertake tasks, including the role of the hippocampus and how it accesses stored information to support these activities.
Connecting the Dots: Neural Mechanisms and Symbiotic AI
So, where do the two lines of research meet? The neural basis provides the detail that is requisite to enable the development of more advanced artificial intelligence systems in the SYMBIOTIK project. In other words, if AI developers know how the brain further leverages information, they will be able to model the AI in a way that would bolster human cognition rather than vice-versa.
For instance, they could incorporate Human-Reference AI models within a Symbiotic AI system to assist users with planning tasks. Picture an AI that retrieves essential information using strategies similar to those employed by the human brain, though in this case, it would not only gather the necessary knowledge but also engage the user in modeling potential outcomes. This approach would enable the AI to generate higher-quality recommendations, as they would be more contextually relevant and aligned with the ongoing decision-making process.
The emphasis on creating a symbiosis within the SYMBIOTIK project aligns with conclusions from recent studies on neuro mechanisms. Both suggest that AI should complement rather than replace human cognitive activity. This approach opens up further opportunities to enhance AI systems in practice, gaining insights into behaviors while staying true to the inherent goals set by human nature.
References
- Jensen, K.T., Hennequin, G. & Mattar, M.G. A recurrent network model of planning explains hippocampal replay and human behavior. Nat Neurosci 27, 1340–1348 (2024). https://doi.org/10.1038/s41593-024-01675-7