Subject: Request for More Neutrality and Precision in AI Responses
Dear OpenAI Team,
I would like to bring attention to an important issue regarding how ChatGPT presents information, particularly in cases where historical events or scientific claims are discussed. While the AI provides detailed responses, it sometimes states things with a level of certainty that may not be fully justified, especially when alternative interpretations or uncertainties exist.
For example, when discussing the Apollo Moon landings, the AI initially framed arguments in a way that strongly favored the mainstream view without properly acknowledging the limits of absolute proof. While the evidence supports the landings, a fair and scientific approach should clearly differentiate between strong supporting evidence and absolute proof, acknowledging the possibility of alternative explanations where reasonable.
Examples of Concern:
- Moon Footprints → The AI originally stated that astronaut footprints on the Moon were “something highly unlikely to be faked by robotic missions.” However, a more neutral phrasing would be: “While footprints strongly suggest human presence due to their depth, shape, and distribution, they do not absolutely prove it, as an unknown robotic system could have theoretically created them.”
- Moon Rock Collection → The AI initially stated that “no such technology existed in the 1960s” to suggest that robots could not have retrieved the Apollo lunar samples. A more accurate phrasing would be: “No such technology is known to have existed in the 1960s, but if such a system did exist, it was not publicly documented.” This revision prevents assumptions about the absolute limits of past technology.
I encourage OpenAI to implement a more cautious approach when presenting widely accepted claims, ensuring that alternative possibilities are acknowledged where appropriate. One possible solution is to adopt wording that explicitly distinguishes between strong evidence and absolute proof, preventing assumptions that could lead to unintentional bias.
Thank you for considering this feedback, and I appreciate the work being done to improve AI-driven discussions.
Best regards,
ong chai hin