Ethical Considerations, Prompt Management and Linguistic Challenges in ChatGPT’s Responses to Controversial Topics: A Comparative Analysis of GPT-3.5 and GPT-4.0 in the Case Study on Sex Robot Influencers

Abstract

This article explores the ethical considerations, prompt management strategies, and linguistic challenges in eliciting responses from large language models, specifically ChatGPT 3.5 and GPT-4.0, on controversial topics. Using the case study of sex robots influencers, the research investigates how these models deal with sensitive and explicit content while adhering to OpenAI’s ethical guidelines. A series of tailored prompts were employed to evaluate and compare the models’ performance, including their ability to provide accurate answers, their adherence to ethical standards, and their linguistic adaptability.

The findings reveal significant differences between ChatGPT 3.5 and GPT-4.0 in their willingness to engage with controversial topics and the quality of their responses. ChatGPT 3.5 demonstrated a more cautious approach, frequently avoiding direct engagement with sensitive content, while GPT-4.0 exhibited a more nuanced understanding but occasionally provided less accurate information. Both models emphasized ethical considerations, redirecting users towards broader discussions on societal and ethical implications.

This study highlights the essential role of prompt management and linguistic adjustments in influencing AI model behavior and demonstrates the limitations and ethical challenges associated with generative AI in addressing controversial subjects. The article concludes with recommendations for future research and the ethical use of AI in complex, sensitive discussions.