Documents and "evidence" are capable of multiple interpretations - including evaluations of reliability and accuracy and authenticity. Furthermore, different documents and evidences can appear to be contradictory - and this contradiction depends on analysis and interpretation - an generally they need to be integrated into a consistent interpretation. This depends on meticulous and careful analysis (including logical, linguistic and stylistic analysis), logical deduction, inference, hypothesis building, hypothesis testing, mutual comparison and evaluation of different hypotheses, historical reconstruction, counterfactual reasoning, decision - together with synthesis and abstraction.
All human intellectual progress (as well as jurisprudence and medicine) are based on a similar process. Scientific papers, books and data-sets are the unit documents and "evidences" which need to be evaluated, interpreted, revised, compared and harmonized. And not enough emphasis is given to the contingent sociological factors that shaped modern mathematics as well as the defeasibility, capacity of improvement, clarification and reformulation of mathematics proofs.
The above constitutes a powerful argument against the extensive use of Wikipedia and social media as vehicles for education and human intellectual progress. But most importantly, it supplies ground for utterly rejecting the use of generative AI, which should be deemed intrinsically worthless and harmful. The subtle process of analysis, interpretation, critical evaluation and harmonization is out of the reach of a crude statistical analysis of tokens extrapolated from already questionable, shallow and largely corrupt, unreliable and compromised data-sets.
No comments:
Post a Comment