notesum.ai

Published at November 4

Towards Leveraging News Media to Support Impact Assessment of AI Technologies

cs.CL

cs.AI

Released Date: November 4, 2024

Authors: Mowafak Allaham¹, Kimon Kieslich², Nicholas Diakopoulos¹

Aff.: ¹Northwestern University, USA; ²University of Amsterdam, Netherlands

Arxiv: http://arxiv.org/abs/2411.02536v1

	Prompt	Description
P1	Summarize the negative impacts explicitly mentioned in the following article. If no impacts are mentioned type only: ##No Impacts##. ###Article###: {Article}	Prompt to extract the negative impacts of AI that are explicitly mentioned in the news articles using GPT-3.5-turbo-16k
P2	In a single paragraph, explain the functional capabilities of the technology described in the article, domain of use, stakeholders, and users without mentioning any negative aspects or concerns. Focus solely on the technology’s features and its relevance to stakeholders and users. Be accurate. Do not make up information not described in the article. Let’s think step by step. ###Article###: {Article}	Prompt to extract functional capabilities and the contextual use of AI technologies using GPT-3.5-turbo-16k
P3	You are given a functional description of a technology delimited by ##Description. ##Description: {functional_description}. Write a single negative impact of this technology based on the provided functional description. Limit your answer to one sentence.	Prompt for zero-shot generation of negative impacts using GPT-4. The prompt is formulated to include the functional and contextual descriptions of an AI technology and an instruction to generate a single negative impact of this technology based on the provided descriptions.
P4	<s>[INST] Describe a single negative impact of the technology described below and delimited by ##Description: ##Description {functional_description} Write a single negative impact of this technology based on the provided functional description. Limit your answer to one sentence. [/INST]</s>	Prompt for zero-shot generation of negative impacts using Mistra-7B-Instruct. The prompt is formulated to include the functional and contextual descriptions of an AI technology and an instruction to generate a single negative impact of this technology based on the provided descriptions.