Scientific Method and Data Science Models

The Problem’s Main Concepts

1. How are DS/ML/AI models equivalent to, or not equivalent to, conservative falsifiable hypothesis testing in “traditional” the scientific method?

2. How might DS/ML/AI models represent a forefront of STEM, extending science and stem into new integrated-STEM areas?

3. In the interplay between:

- Probability & Statistics

- DS/AI/ML

- Falsifiability & Hypothesis Testing

- Nonlinearity

3.1 How do all of these fit together in this context?

3.2 Where exactly is: ‘the scientific method’?

4. Has the generalized-STEM question been resolved within a timeline of western science and the timeline of the development and refinement of the hypothesis testing, falsifiability, and “the scientific method”?

Overview

1. People would like this topic to be:

- clean & clear

- finite and small

- resolved and confirmable

2. But, and I am optimistic about this, this topic appears to be:

- fragmented

- large, broad, and with open-ended scope and potential

- unresolved and with an unknown overall shape

in such a way that there is significant room for expansion into new STEM-integrated, structure-function-ratcheting tools and methods.

Speculation

DS/AI/ML models are surprisingly equivalent in many cases to the preferred linear model tests for conservative hypothesis testing, which may catalyze a significant expansion in applications of enlightenment-STEM tools, and represent perhaps a new era and conception of how the enlightenment valuation and utilization of ~reason (roughly as expounded by advocates such as Steven Pinker in works like “Enlightenment Now”).

About The Series

This mini-article is part of a series to support clear discussions about Artificial Intelligence (AI-ML). A more in-depth discussion and framework proposal is available in this github repo:

https://github.com/lineality/object_relationship_spaces_ai_ml