The Nooscope manifested: AI as instrument of knowledge extractivism

Area of Science:

• Artificial Intelligence
• Machine Learning
• Sociology of Technology

Background:

• The project aims to mechanize reason through machine learning.
• Machine learning involves data, algorithms, and models, with training datasets reflecting social origins of machine intelligence.
• The history of AI is characterized by the automation of perception.

Purpose of the Study:

• To provide enlightenment on the project to mechanize reason.
• To analyze the components of the machine learning assembly line.
• To discuss the societal implications and limitations of current AI models.

Main Methods:

• Examining the structure of machine learning: data, algorithm, model.
• Analyzing the role of training datasets and their social origins.
• Reviewing the history of AI as the automation of perception.
• Understanding learning algorithms as methods for compressing the world into statistical models.

Main Results:

• Machine learning operates as an assembly line: data, algorithm, model.
• Training datasets reveal the social origins of machine intelligence.
• AI's history is rooted in automating perception.
• Learning algorithms create statistical models, which are useful but imperfect, particularly in detecting novelty.
• Adversarial intelligence presents a challenge to statistical intelligence in the context of labor.

Conclusions:

• Statistical models, while useful, inherently fail to detect the new.
• The societal impact of AI, particularly concerning labor and the rise of classification and prediction bots, requires careful consideration.
• A distinction is drawn between adversarial and statistical intelligence in the age of AI.