Domain Driven Reasoning on ChatGPT Trust

• Most people are not interested in formalism since they don't know much about it.
• Most people are super interested in Artificial Intelligence even they don't know much about it.

If you think my comments make any sense, bear with me, I have more to say.

• For most people, we don't need to know much about Artificial Intelligence to use it.
• However, for most people, we need to know formalism in order to use Artificial Intelligence well.

The last sentence is my key point.

If one knows Plato's philosophy, he postulates that there is a concept of the Eternally True that lies beneath the seemingly mutable world. Plato believed in the World of Ideas, which is a collection of "molds" or "forms" that are eternal and unchanging, and everything in the material world is created in accordance with them.

According to Plato, true knowledge can only be acquired through our reason. Akin to the way everyone arrives at the same conclusion when solving a mathematical expression, reason is universal and everlasting because it only conveys perpetual and universal concepts.

Suppose, that you can leverage ChatGPT to easily manufacture a lot of quasi-knowledge, those quasi-knowledge would take the form of the "knowledge" that makes descriptions, judgements, pitches, ideologies, and you cannot reject them based on their appearance because their appearance is too "valid." Instead, you have to accept / reject them by reason.

And the reason for ChatGPT trust verification can be carried out by domain-driven design. This is a process to model the domain specific semantics and concepts and apply rules to define the fundamental constraints at work that ensures the knowledge's trustworthiness.

Detailed discussion see my last week's blog post

Please subscribe my substack for more in-depth thinking!