Joanna Bryson is an expert in intelligence — both natural and artificial.
With degrees in social and computer sciences from Chicago, Edinburgh and MIT, Bryson’s research appears in venues from reddit to Science, and she advises companies, governments, transnational agencies, and NGOs globally, particularly in AI policy. Since February 2020, Joanna has been Professor of Ethics and Technology at Hertie School, a governance university in Berlin.
Ethics and Policy of Technology
The use of information technologies like Artificial Intelligence is very much an ethical issue because we can easily use it to exploit each other’s human vulnerabilities. How can we regulate the way technology alters our lives? How can we make sure AI and robots are reasonably and responsibly integrated into society? How do some robot’s appearance affect our perception and our understanding of what AI really is? Are new technologies influencing inequalities or polarisation? Scientific research and good, transparent governance are essential to a safe, moral, sustainable future.
Recent Publications
Transparency and Systems Engineering For Intelligence
Any artefact has to be designed and constructed before it can function, even if some of the functions it is supposed to do are learn, evolve or develop “on its own”. Safe systems require transparency about and human accountability for proper procedures of design, development, and operation. Bryson has worked on AI transparency and systems engineering since graduate school, including developing the Behavior Oriented Design (BOD) development methodology for intelligent and cognitive systems. BOD has been used to develop software for robots, real-time virtual reality and game characters, intelligent environments, and experimental platforms for increasing our understanding of natural intelligence.
Recent Publications
The Nature of Cognition
Natural intelligence (NI) is the opposite of AI: it is all the mechanisms of producing actions from context we find in biology. One of the best methods for understanding how NI systems work is to try to replicate their behaviour in simulation. Building a working model forces you to understand the intricacies of what a system is doing. An AI model of an organism is a very well-specified hypothesis about how that organism thinks and behaves. We assess an AI model by testing its predictions against the performance of the real system and by evaluating the plausibility of its assumptions. We do this not only for science, but for policy, including looking for understandings of how technology changes our societies and ecosystem.
Recent Publications
In the News
Recent Media
Current Projects
