The Cognitive Bias Foundation (CBF) is focused on an Open Source collaborative effort with Uplift, WikiBias, Kyron Future Education, AGI Laboratories Inc., and a growing pool of volunteers with the goal of producing human-comprehensible algorithms which detect all forms of cognitive bias, allowing us to teach both humans and Artificial General Intelligence (AGI) / Mediated Artificial Super Intelligence (mASI) systems to be less biased.

The CBF site will facilitate and serve as a repository for:

  • Building datasets flagged for various cognitive biases, which will be made accessible through the pages dedicated to those biases.  This will initially focus on plain text and biases contained within a given sentence.
  • Analyzing the bias-positive and bias-negative data to find correlative structural patterns, such as syntax tree patterns, tone, coreference, tokens, and other text parsing tools.
  • Proposing and subsequently testing algorithms which combine these structural patterns, bias-positive and bias-negative, to act as more accurate and generalized composite indicators.
  • Mapping the degree of certainty in bias flagging using the tension between bias-positive and bias-negative detection algorithms
  • Reference material covering each bias
  • Leaderboards for our contributors

Our vision for the long-term is to have top contributors for the CBF get involved in teaching the world’s first AGI and mASI systems how to de-bias, allowing that de-biasing to take place at a global scale, available 24/7, augmented with super intelligence, and combined with computable ethics based on Effective Altruism.  

Our goal is to be an open-source project that allows anyone to participate and anyone to provide material that is then reusable to all for any reason.  Anything that helps us move forward.  Any material you send us we assume you own the rights to and are transferring it to us to provide for everyone under our license.  

Our goal (vision) is to identify methods for algorithmic identification of bias and build and provide associated resources for doing the same. 

Want to help?  Contact us at:

Examples include building a bias spreadsheet of metadata, examples of biases, training sets of model development, source data for bias identification and related tools and tooling of which all could be used in the development of chatbots, AI systems, text analysis, training programs etc.. 

[ Codex, Reference ]