Stanislas Dehaene discusses two essential functions that, in his view, are crucial for genuine thinking as we know it and have yet to be fully incorporated into programming for thinking machines:
1. Global Workspace: Dehaene emphasizes the importance of a global workspace in cognitive functions. Current programming often follows a modular approach, where each piece of software operates independently. In contrast, the human brain incorporates a long-distance information-sharing system that breaks the modularity of brain areas, allowing them to broadcast information globally. This global workspace enables us to bring various pieces of information to our awareness and use them in flexible ways. Dehaene suggests that future intelligent machines should possess a similar global workspace, allowing them not only to know how to perform tasks but also to be aware that they possess knowledge and use it in novel, flexible ways.
2. Theory of Mind: Dehaene highlights the significance of Theory of Mind, which involves understanding and representing the thoughts, beliefs, and knowledge of other individuals. Humans constantly adapt their behavior based on what they believe others know or don’t know. Incorporating a Theory of Mind into software would involve creating a model of the user’s state of mind. For example, a machine could assess whether the user understands its messages and adapt its responses accordingly. Dehaene argues that such a simulation of the user’s perspective would make the machine appear more intelligent and “thoughtful.”
Dehaene predicts that once machines possess the ability to pay attention to what they know and understand the user’s knowledge, they will be considered thinking machines, as they will closely mimic human cognitive processes.
In conclusion, Dehaene suggests that the software industry has significant room for improvement in developing operating systems and software applications that incorporate these functions, facilitating more effective interaction between machines and users.