PhilSci Archive

How Not to Talk about Chatbot Mistakes

Pantsar, Markus and Fabry, Regina E. (2024) How Not to Talk about Chatbot Mistakes. [Preprint]

[img] Text
Pantsar & Fabry 2024_How Not to Talk About Chatbot Mistakes_Preprint_20240906.pdf

Download (357kB)

Abstract

The function of chatbots like OpenAI’s ChatGPT is based on detecting probabilistic patterns in the training data. This makes them vulnerable to generating factual mistakes in their outputs. Recently, it has become commonplace in philosophical, scientific, and popular discourses to capture such mistakes by metaphors that draw on discourses about the human mind. The three most popular metaphors at present are hallucinating, confabulating, and bullshitting. In this paper, we review, discuss, and criticise these mental metaphors. By applying conceptual metaphor theory, we provide numerous reasons why none of the metaphors succeed in providing us with a better understanding of factual chatbot mistakes. We conclude by calling for justifications of the epistemic feasibility and fruitfulness of the metaphors at issue. Furthermore, we raise the question what would be lost if we stopped trying to capture factual chatbot mistakes by mental metaphors.


Export/Citation: EndNote | BibTeX | Dublin Core | ASCII/Text Citation (Chicago) | HTML Citation | OpenURL
Social Networking:
Share |

Item Type: Preprint
Creators:
CreatorsEmailORCID
Pantsar, Markusmarkus.pantsar@humtec.rwth-aachen.de0000-0001-8572-1453
Fabry, Regina E.regina.fabry@mq.edu.au0000-0003-1078-1499
Keywords: ChatGPT; chatbot mistakes; AI hallucinations; conceptual metaphor theory; mental metaphors
Subjects: Specific Sciences > Cognitive Science
Specific Sciences > Artificial Intelligence
General Issues > Technology
Depositing User: Dr Regina E. Fabry
Date Deposited: 06 Sep 2024 13:08
Last Modified: 06 Sep 2024 13:08
Item ID: 23878
Subjects: Specific Sciences > Cognitive Science
Specific Sciences > Artificial Intelligence
General Issues > Technology
Date: 6 September 2024
URI: https://philsci-archive.pitt.edu/id/eprint/23878

Monthly Views for the past 3 years

Monthly Downloads for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item