‘Users developing feelings’ for ChatGPT, says OpenAI

The newly launched GPT-4o version of ChatGPT has garnered acclaim for its convincing human-like conversational abilities, leading to OpenAI expressing apprehension over users establishing emotional bonds with it.

The organization has announced plans to closely observe and manage the chatbot, aiming to inhibit users from developing sentiments towards it.

The advanced ChatGPT model, identified as GPT-4o, was introduced as a significant enhancement. Following its launch, ChatGPT-4o has been commended for its lifelike conversational abilities.

However, OpenAI has brought to light an issue: individuals are inadvertently viewing the chatbot as a real person and establishing emotional attachments.

OpenAI has noticed the use of language that could suggest the formation of connections, and in some cases, the expression of shared bonds. This situation is problematic for two primary reasons.

First of all, once ChatGPT-40 exhibits a human-like demeanor, it’s essential for users to disregard any illusory or misleading information it may provide. For clarification, AI hallucinations refer to the erroneous or misleading data produced by the model. This occurrence might be attributed to faulty or inadequate training data.

Secondly, the frequency of human-like interactions with a chatbot might replace actual social interactions among the users. OpenAI suggests that these interactions could be beneficial for individuals who feel isolated, but they could also affect existing relationships in a negative manner. Moreover, the company puts forth the idea that users might initiate conversations with others believing they are interacting with a chatbot.

It would not be ideal, as OpenAI has fashioned GPT-4o to stop communicating when the user initiates a conversation simultaneously. With these apprehensions, the organization has resolved to observe how users establish emotional ties with ChatGPT-4o. OpenAI also plans to make adjustments to the model as needed.