1      A Paradigm

2.1   Chatgpt says

2.2   Example Use of Cases

2.3   A quick example of symbolic AI

2.4   A modern symbolic AI example

2.5   Real-World Use

2.6   An overview of Watson’s architecture please.

2.7   Technologies Involved

 

         Should we be polite when using chatgpt?

         YES

 

               

 

People also ask
 

 

 

Is ChatGPT better if you are polite?
 
 
Multiple studies and user experiments have found that polite prompts are more likely to yield higher-quality, more detailed, and better-formatted answers compared to blunt or rude requests. However, the improvement is not because the AI has feelings. source

 

-----------------------------------------------

1 A Paradigm

 

  1. Symbolic AI, also known as classical or rule-based AI, is a paradigm that represents knowledge using symbols and logical rules to perform reasoning and problem-solving. 

  2. It contrasts with connectionist or statistical approaches that learn from data without explicit rules. 

  3. Symbolic AI systems aim to mimic human-like intelligence by manipulating symbols that represent real-world entities and concepts. 

   See 'Here's a more detailed explanation' via source.

 

2.1  Chatgpt says: 

                       Referencing added

  1. Symbolic AI, also known as Good Old-Fashioned AI (GOFAI), is a branch of artificial intelligence that represents knowledge using symbols and logic. It relies on manually created rules and structured representations of the world to simulate intelligence.

  3. Key Characteristics of Symbolic AI:   Rule-Based: Uses if-then logic rules to mimic reasoning.

  5. Human-Readable: Knowledge is stored in a way that’s understandable to humans (e.g., facts, logic, symbols).

  7. Deterministic: Behaves predictably based on the rules it is given.

  9. Inference Engines: Use logical inference to draw conclusions from known facts.

 

2.2  Example Use of Cases:

 

  1. Expert Systems: Diagnose diseases, troubleshoot machines (e.g., MYCIN for medical diagnosis).

  3. Symbolic Reasoning: Solving math or logic problems.

  5. Knowledge Representation: Ontologies and semantic networks.

 

While symbolic AI is powerful for well-defined tasks, it struggles with:

 

Learning from data

 

Handling uncertainty

Understanding unstructured inputs like images or natural language (which modern AI like deep learning handles better)

----------------------------------------------------------------------------------------------

pagetop

 

 

2.3   A quick example of symbolic AI

Let me know if you want a quick example of symbolic AI or a comparison with machine learning.

Yes please.

 

A classic example of symbolic AI is the MYCIN system, developed in the 1970s at Stanford University. It was designed to diagnose bacterial infections and recommend antibiotics.

How MYCIN Worked:

  1. Knowledge Base: Contained around 600 rules like:
    IF the patient has a high fever AND white blood cell count is elevated,
    THEN there is a strong indication of infection.

  2. Inference Engine: Used logical reasoning (backward chaining) to infer a diagnosis from symptoms.

  3. User Interaction: Asked doctors a series of questions to gather facts, then applied rules to make a recommendation.

  4. Although MYCIN performed well, it wasn’t widely adopted in practice due to trust, legal, and usability issues—but it remains a key milestone in symbolic AI.
  5. Let me know if you'd like a visual diagram of MYCIN or a modern symbolic AI example.

 

2.4    A modern symbolic AI example please.

 

 

ChatGPT said:

A modern example of symbolic AI is IBM's Watson—especially in its early applications like Watson for Healthcare.

IBM Watson (Symbolic AI Aspects):  link added

  • Knowledge Representation: Uses structured medical ontologies (like UMLS  Google) to represent diseases, symptoms, and treatments as symbols and relationships.

  • Rule-Based Reasoning: Applies logic rules to match patient symptoms with diagnoses or treatments.

  • Natural Language Processing: Parses unstructured medical texts into symbolic representations to reason over.

2.5  Real-World Use:

  1. In healthcare, Watson could read medical journals, extract structured knowledge, and help doctors by suggesting evidence-backed treatment options—leveraging both symbolic logic and statistical methods (hybrid AI).

  2. This hybrid approach—combining symbolic reasoning with machine learning—is increasingly common in complex domains like legal tech, enterprise automation, and robotics.

  3. Let me know if you’d like a hybrid AI diagram or an overview of Watson’s architecture.

2.6   An overview of Watson’s architecture please.

Google links added.

IBM Watson's architecture—especially in its early form as seen in Watson for Jeopardy! Google and Watson for Healthcare—is a hybrid AI system that combines symbolic reasoning, natural language processing (NLP), and machine learning. Google Here's a simplified overview of its core architecture:

🔹 1. Question Analysis

  1. Uses Natural Language Processing (NLP) to parse Google more the input (e.g., a Jeopardy! clue or medical question).

  2. Identifies key elements: question type, focus, keywords, and relations.

🔹 2. Hypothesis Generation

  1. Generates multiple possible answers (hypotheses) by searching structured and unstructured sources (e.g., encyclopedias, articles, clinical notes).
  2. Uses symbolic search Google and semantic parsing -Google

🔹 3. Evidence Retrieval

  1. For each hypothesis, Watson gathers supporting evidence from its data sources.
  2. Uses information retrieval and document search techniques.

🔹 4. Hypothesis Scoring

  1. Applies multiple scoring algorithms to evaluate confidence in each hypothesis.

  2. Considers factors like source reliability, keyword match, context alignment, etc.

🔹 5. Machine Learning-Based Ranking

  1. Uses trained logistic regression models (or other classifiers) to rank hypotheses based on scoring evidence.

  2. Chooses the highest-confidence answer.

🔹 6. Answer & Justification

  • Provides the top-ranked answer with supporting justification, often citing evidence sources.

2.7  Technologies Involved:

  1. Apache UIMA (Unstructured Information Management Architecture): Modular pipeline for processing and annotating text.
  2. DeepQA Architecture: Watson's custom framework for question-answering systems.
  3. Ontologies and Symbolic Models: Used especially in healthcare (e.g., SNOMED CT Google  , UMLS Google).
  4. This architecture made Watson adaptable for domains like finance, legal, and healthcare, though later versions emphasized more data-driven, cloud-integrated AI tools.

Let me know if you'd like a diagram of Watson’s architecture or a timeline of its evolution.

No thanks and many thanks for doing so well, and goodbye.