IARPA wants to improve human/machine forecasting
- By Kevin McCaney
- Feb 01, 2016
The Intelligence Advanced Research Projects Activity is looking to improve it intelligence forecasting by optimizing the combination of human and machine interactions with regard to geopolitical and geoeconomic events.
IARPA’s Hybrid Forecasting Competition is expected to be a multidisciplinary effort involving academia and industry to find the best ways that human judgements can be combined with computer-generated data, according to an IARPA release. The agency is holding a proposer’s day Feb. 3 to gather information in advance of a solicitation for the program.
In its announcement, IARPA notes that analysts have struggled with how to best combine and weight machine-based forecasts with human judgements. As part of the HFC program, the agency wants to develop:
- Protocols that train human forecasters to optimally combine and weight human and machine judgments and forecasts
- New predictive models that incorporate both machine data and human judgments
- Algorithmic forecasting agents that interact with human forecasters/forecasts inside crowdsourced forecasting platforms.
Although IARPA singled out those three areas, it said proposal don’t have to be limited to them.
Among the disciplines the program expects to draw from are social and behavioral sciences; operational forecasting disciplines covering, for example finance, macroeconomics, meteorology and geopolitics; computer science and software development; and methodologies in statistics, psychometrics, polimetrics and econometrics.
Kevin McCaney is a former editor of Defense Systems and GCN.