Epistemology’s Methodism and Particularism and Similarities to Knowledge Engineering Principles
by Wei Jing HO – Sunday, 22 September 2024, 2:03 AM
Number of replies: 4
In answer to 2.4 The problem of the criterion. During the reading and viewing of the video, it made me think of the similarities behind certain Computer Science and Knowledge Engineering Principles.
First to sum-up the distinction between methodism and particularism
- Methodism: Utilises general rules for acquiring knowledge and then determines if certain beliefs qualify as knowledge.
- Particularism: Assumes some specific belief are knowledge and seeks to infer general sets of rules.
————————————-
To share a quote by Edward Feigenbaum of Stanford University in regards to using Knowledge Engineering to build intelligent systems:
“The first principle of knowledge engineering is that the problem-solving power exhibited by an intelligent agent‘s performance is primarily the consequence of its knowledge base, and only secondarily a consequence of the inference method employed.
Expert systems must be knowledge-rich even if they are methods-poor. This is an important result and one that has only recently become well understood in AI.
For a long time AI has focused its attention almost exclusively on the development of clever inference methods;…. The power resides in the knowledge.”
One of his paper is shared here. Ref: https://stacks.stanford.edu/file/druid:qy055zd8682/qy055zd8682.pdf
————————————-
In the beginnings of AI history, research began with creating complex inference methods which might draw similarities with the methodism‘s assertion that one needs a general method or criteria for determining what counts as knowledge. Such exploration resulted in the rule-based systems (Expert Systems), deductive logic systems, theorem provers which were created with constructing formal logic and deductive reasoning.
- Forward Chaining applies inference rules to known facts so that the system can derive new facts.
- Backward Chaining starts with a goal and work backwards to check if the known facts support the goal.
While they are less dominant now in AI system designs, they have certain influence in the creation of programming logic in Computer Science e.g., Prolog, SQL, and are part of the later process of creating more efficient and streamline AI systems.
————————————-
In our current timeline, AI advancement leans towards Machine Learning, Deep Learning and utilisation of probabilistic methods that uses large amounts of data to train an AI model, creating systems that learns from data. This is more align with Particularism‘s assertion that they know what is knowledge without a method for finding what knowledge is.
————————————-
When I refer back to the question of which view (Methodism vs. Particularism) is more plausible, I will say both are plausible, both are equally important.
Like blind men exploring different parts of an elephant, each explorer derived a viewpoint that does not encapsulate the whole elephant. Each explorer shines their own light into their segment of darkness, all of their sensemaking and insights are valuable.
And like AI system designs, while the world is moving towards AI systems trained on large amounts of data, one should note that:
-> as such AI systems learn from data… there will be certain formation of “rules” and “methods” during their learning process
-> the rules and methods of the data formulated from the learning cycles can then be used to create smaller, more efficient, less heavy and able to run on edge devices (Edge AIs)… as the AI designer can revert back to the older knowledge engineering designs to blend in inference type systems.
Depending on the need, similar to the considerations behind the designs of intelligent systems, both the teachings of Methodism and Particularism will be useful.
Re: Epistemology’s Methodism and Particularism and Similarities to Knowledge Engineering Principles
by James Carmichael – Sunday, 22 September 2024, 9:40 PM
Thanks, Wei Jing. I’d been aware of this distinction and the recent shift at a high-level, reader-of-The-Economist level. This explanation of yours was enriching and added to that; thanks for taking the time to share it.
Re: Epistemology’s Methodism and Particularism and Similarities to Knowledge Engineering Principles
by Ana Selimbegovic – Monday, 23 September 2024, 10:07 PM
A fantastically thoughtful and pertinent post, Wei, thanks so much for this. I also agree with everything you’ve said, continuously evaluating and enhancing both approaches is necessary for meaningful development not just of our methods of inquiry, but civilization itself.
Re: Epistemology’s Methodism and Particularism and Similarities to Knowledge Engineering Principles
by David Laflamme – Tuesday, 24 September 2024, 12:41 AM
Great post Wei Jing! I liked and learned a lot from your explanation and analysis of the problem of criterion. What came to my mind as I was reading your great examples and analogies is that in the case of AI there is an objective standard/truth – the human creator’s particularism and methodism. This in no way detracts from your point, but in my view, reinforces the need to explore our challenges of understanding knowledge.
Re: Epistemology’s Methodism and Particularism and Similarities to Knowledge Engineering Principles
by Kimberly Inge – Saturday, 28 September 2024, 10:29 PM
Thank you for your post, Wei Jing. Although I don’t have much experience or deep understanding of AI and engineering systems, I’ve had some recent experiences with AI, particularly a ride in a Waymo autonomous taxi and using Chat GPT to create a rubric.
I’m not sure what I’m about to say completly relates to your astute explanations and observations, but I’ll put the idea on the table anyway. Thinking about what we’re focusing on this week and what you wrote–“In our current timeline, AI advancement leans towards Machine Learning, Deep Learning and utilisation of probabilistic methods that uses large amounts of data to train an AI model, creating systems that learns from data. This is more align with Particularism’s assertion that they know what is knowledge without a method for finding what knowledge is.” I’m starting to think we need another category apart from knowing what knowledge is (particularism) and the method/criteria of knowledge (methodism)…perhaps we need a term that captures DOING knowledge . With missteps along the way, Waymo and Chat GPT seem to be doing knowledge quite well in my estimation.
Kimberly