Using AI in Applications

AI is about using conceptual modeling and reasoning to solve cognitive problems. This is more than using mathmatical models to solve physical problems (like probabalistic simulations of the environmental consequences of nuclear power plant failure) or computers doing organization tasks (like managing corporate accounting or sales force automation).

Cognitive processes include not only traditional activities like planning and decision-making but building systems that can learning things not built in by their developers, can explain how they work and reflect on and justify their actions and that can represent beliefs, desires, and intentions.

The NEWFOL Framework is software that can be customized to add symbolic AI functionality to existing applications.

Symbolic AI vs Neural Nets

Symbolic AI as well as the use of neural nets in LLMs uses ideas that were well discussed 50 year ago when the most powerful computers were less powerful than today's smartphones. What generative AI and LLMs has taught us is that massively scaling old algorithms (e.g., neural nets and tenor algebra) using the latest hardware make a qualitive difference in functionality.

A second take away from current LLM usage is that without a grounding in hard facts, the results produced by LLMs are hard to validate and need to be taken with a grain of salt—especially in life-critical applications.

This suggests another look at symbolic AI. The NEWFOL Framework interprets logic computationally rather than set theoretically and offers a new form of symbolic AI that uses the analysis of logicians about how thinking works and at the same time handles some of the difficulties inherent in generative AI.

The NEWFOL Framework

The NEWFOL framework is a rich development environment for adding cognitive capabiities to an applications. It includes: tools for developing models and connecting them to new code, exiting application code, and programs that can communicate with people or use sensors and affectors to understand the world; and linkable runtime libraries and network accessable APIs for deployment.

The framework focuses on modeling thinking about an application and reliably useing knowledge of the world and logic (and possibly LLMs) to solve problems.

IBUKI full range of IBUKI products that supplement the NEWFOL Framework. include: the NEWFOL Framework with its associated runtimes; prebuilt 'mental' models including both 'common ground' knowledge and specialized subjects knowledge.; and many useful subsystems for building AI apprlications.