Frequently Asked Questions
- What is narrative computation?
- Narrative computation is an attempt at creating AI systems that are accurate and explainable.
- What do accurate and explainable mean?
- Accuracy refers to the tendency for an AI system to not provide incorrect information and make best-effort attempts at solving problems, especially in new or unexpected situations;
- Explainability refers to the tendency for an AI system’s behaviors to be usefully-explained in human terms; i.e. not a black box.
- Why are accuracy and explainability important?
- Accuracy is important for critical applications where wrong answers may have serious consequences, like engineering, healthcare, autonomous driving, and many more; especially involving difficult edge cases that must be handled in safely and reasonably.
- Explainability is important because understanding how AI systems behave is critical for fixing problems, for improving performance and for increasing confidence in the design of the systems.
- How is the narrative computation approach to AI different from generative AI approaches, like ChatGPT?
- Narrative computation is a form of symbolic approach to AI (relying on systems of rules to create models of the world); whereas generative AI is a form of neural network approach to AI (relying on patterns that emerge in large amounts of data);
- Although symbolic approaches and neural network approaches can be combined, narrative computation is a symbolic AI approach that does not utilize neural networks directly;
- As such, narrative computation is a very different approach to AI from generative AI approaches.
- How do narrative computation systems perform compared against generative AI systems?
- NX-RESEARCH’s systems are still in an extremely early research stage; a comparison has not been published.
- Given that there is no published performance comparison against generative AI systems, what makes narrative computation a promising line of research?
- At NX-RESEARCH, we believe that accuracy, robustness, and explainability will become increasingly vital differentiators in critical AI applications, like engineering, healthcare, autonomous driving, and many more;
- And we believe that the narrative computation approach offers the possibility of systems that may someday be more accurate, robust and explainable than generative AI systems.
- How long before a performance comparison will be available for narrative computation systems against generative AI systems?
- NX-RESEARCH’s systems are still in an extremely early research stage; our current rough estimate for some early comparisons and demos is within about 18 months.
- What are some of the key research milestones or obstacles for deploying narrative computation systems?
- NX-RESEARCH’s systems are still in an extremely early research stage; we haven’t and don’t intend to publish any public roadmaps at this time.
Further reading:
Introduction to Narrative Computation (White Paper)
Job Posting: Software Engineer – AI Evaluation