Decision Intelligence & Neurosymbolic AI –
A guide for leaders
Dr. Lorien Pratt’s 2008 vision of “Decision Intelligence” has come to fruition with Rainbird’s emerging neurosymbolic capabilities. This joint whitepaper, with Rainbird co-founders James Duez and Ben Taylor, explores the potential of neurosymbolic AI delivering Decision Intelligence.
Abstract
The last 12 months have seen significant breakthroughs in the advancement of AI.
This paper explains the potential of Decision Intelligence and Neurosymbolic AI in maximising the value that can be extracted from AI today in the pursuit of digital transformation, while minimising risks.
It addresses key questions and concerns being voiced by leaders about the use of these technologies. In particular, in the context of Large Language Models (LLMs) like GPT-4, which are powerful creators of generated content but lack reasoning, logic and explainability.
We introduce a neurosymbolic approach, which integrates extended knowledge graphs, LLMs and neural networks to deliver the combined strengths of these technologies while avoiding their weaknesses. It discusses the ability of this composite approach to address the limitations of current AI with a focus on improving explainability, reliability and trust.
We highlight the benefits of Neurosymbolic AI, including its ability to handle the decision complexity that is more typical of current enterprise demands, reduced data requirements in a world that has had a saturated “data first” mindset, and provide transparency and explainability where governments are making a concerted effort to introduce greater governance obligations to protect the population at large.
Finally, we outline Rainbird’s specific AI lifecycle. This includes the steps of; building models from documentation and expertise, refining them with data, testing and deploying them, interacting with them in natural language, and ensuring they learn from experience.