Presenter Information

Mariah Jacobs, Pacific University

Location

Tacoma

Event Website

https://www.pugetsound.edu/academics/departments-and-programs/undergraduate/philosophy/philosophyconference/

Start Date

1-2-2019 12:00 PM

End Date

1-2-2019 12:50 PM

Description

The influence Karl Popper’s falsificationist model has had on the scientific method and the demarcation problem is troublesome for the field of artificial intelligence (AI). According to Popper, the falsifiability of a hypothesis is a necessary condition for its scientific validity. Because the falsificationist model has been formative in the development of modern philosophy of science, it has become the primary way in which we demarcate the scientific from the non-scientific. However, as a consequence of our current, limited understanding of mental properties—such as intelligence, thought, and personal identity—I argue that it is unclear whether hypotheses concerning the design of artificial intelligence—particularly strong AI—are truly falsifiable. If this is the case, society’s approach to and attitude towards future AI research and development regarding scientific methodology is in need of reevaluation. I conclude that we should either (1) aim to better define the concepts in philosophy of mind that we attribute to artificial intelligence to understand how they can be falsified in order to make quicker and more meaningful progress in AI or (2) recognize that there are aspects of AI that simply cannot be falsified and adapt our current scientific methodology to something more appropriate and inclusive than Popper’s falsificationist principles.

Type

event

Included in

Philosophy Commons

Share

COinS
 
Feb 1st, 12:00 PM Feb 1st, 12:50 PM

The Truth in the Falsification of Artificial Intelligence

Tacoma

The influence Karl Popper’s falsificationist model has had on the scientific method and the demarcation problem is troublesome for the field of artificial intelligence (AI). According to Popper, the falsifiability of a hypothesis is a necessary condition for its scientific validity. Because the falsificationist model has been formative in the development of modern philosophy of science, it has become the primary way in which we demarcate the scientific from the non-scientific. However, as a consequence of our current, limited understanding of mental properties—such as intelligence, thought, and personal identity—I argue that it is unclear whether hypotheses concerning the design of artificial intelligence—particularly strong AI—are truly falsifiable. If this is the case, society’s approach to and attitude towards future AI research and development regarding scientific methodology is in need of reevaluation. I conclude that we should either (1) aim to better define the concepts in philosophy of mind that we attribute to artificial intelligence to understand how they can be falsified in order to make quicker and more meaningful progress in AI or (2) recognize that there are aspects of AI that simply cannot be falsified and adapt our current scientific methodology to something more appropriate and inclusive than Popper’s falsificationist principles.

https://soundideas.pugetsound.edu/psupc/psupc2019/friday/4