What are you working on? Checking in with Frederic Moisan
Hi, I’m Fred. I am a behavioural economist with a background in artificial intelligence and computer engineering. At Fetch, I apply economic theories to the design of decentralised marketplaces where economic agents extract value from interacting with one another. Multiple issues arise from this, many of which are associated with the nature of the agent population. While economics traditionally deals exclusively with human actors, Fetch extends this domain to a new world inhabited by a diverse crowd of artificial autonomous agents. These agents have potentially high heterogeneity in terms of computational ability, intelligence, knowledge, beliefs and motivations. My research focuses on designing economic incentives that support efficient and fair interactions between agents. The role of such incentives is to attract honest agents who can make relevant, meaningful contributions, while preventing others with malicious intentions from profiting by manipulating the network. I combine various tools originating from economics and computer science, including game theoretic analysis, agent-based modelling, machine learning methods, experiments, and simulations.
At Fetch we seek to promote the emergence of collective intelligence through the aggregation of individual knowledge and opinions. In order to achieve this, we are designing markets that incentivise agents to truthfully reveal their best individual prediction regarding specific outcomes. For example, suppose you wish to know the likelihood that it will rain tomorrow. You may then create a market that any other agent (e.g. weather station agents) can use to submit their prediction. Upon the outcome being discovered tomorrow, you will then reward all the bidding agents proportionally based on the accuracy of their prediction. While such prediction markets are already being used in specific real world contexts (e.g. weather forecasts), at Fetch, we wish to democratise the technology so that any single agent can query the crowd at any time about any information of interest. The resulting decentralised marketplace will extensively exploit the capability of the smart contract technology available on the Fetch blockchain. This leads to renewed challenges as a wide variety of complex queries can then be asked. These include the prediction of unobservable or hypothetical events (‘how much traffic congestion would there be in the event of a bus strike?’), subjective judgements (‘how likely is a car accident on this road?’), or outcomes sensitive to the market maker’s behaviour (‘how long will it take me to drive to my destination?’). The nature of such queries make it significantly more difficult to design suitable incentives.
The quality of the wisdom of the crowd strongly relies upon the independence and diversity of individual unbiased opinions. These can be in short supply in human society. Similar biases can be expected in a world of artificial agents that can freely interact and influence each other. This problem becomes increasingly more complex when interactions involve a mixture of both human and artificial actors, as is supported by the Fetch network.
Exploring the intersection of artificial intelligence and collective intelligence represents an exciting new area of research that will undoubtedly be a topic of increasing focus over the coming years.
Such a research programme is highly interdisciplinary as it touches on recent advances in social sciences, economics, computer science, and network science. At Fetch, our goal is to develop a controlled environment that can allow collective intelligence to thrive. We welcome all kinds of collaborations with talented researchers who have relevant backgrounds and expertise.
