A naukri.com initiative
Towards Data Science
1w
348
Image Credit: Towards Data Science
The struggle of Artificially Imitated Intelligence in specialist domains
- Artificially Imitated Intelligence (LLMs) appears to be struggling in specialized domains like finance, medicine, and law due to its limited ability to reason.
- Ontology is a knowledge representation of the domain that could help LLMs overcome its weaknesses and add semantic understanding about how things work.
- Ontologies can ensure complete audit trails, consistency, and objective interpretations in specialist areas like banking regulations, asset and liabilities management with minimal effort.
- LLMs are already making an impact on customer service chats and reducing human costs but is still prone to hallucination.
- Moreover, LLMs can leverage knowledge graphs to achieve better results in Retrieval-Augmented Generation (RAG).
- Ontologies and Knowledge Graphs are game-changers for real intelligence and for future specialist AIs that reduce hallucinations.
- An example of a simple ontology created with the CogniPy library shows how structure, data, and logic can be united with minimal effort.
- The example was able to evaluate rules, make deductions, and explain with complete audit trails.
- Ontologies could benefit businesses for enforcing a common dictionary, understanding the rules, reducing effort on interpretation, and adding more knowledge to specialist fields.
- A proof of concept that unites structure, data, and logic in a smart data structure, both human and machine-readable, could claim the title of AI.
Read Full Article
20 Likes
For uninterrupted reading, download the app