OpenAI Warns ChatGPT Voice Mode Users May Develop ‘Social Relationships’ With AI
In early testing, OpenAI noticed some users connecting using ChatGPT’s Voice Mode.
On Thursday, OpenAI cautioned that ChatGPT’s Voice Mode functionality might lead to users socializing with the AI model. The system card for GPT-4o, which the business evaluated and studied, described the AI model’s dangers and protections. People might anthropize and bond to the chatbot, among other problems. The danger was increased when early testing showed it.
ChatGPT Voice Mode May Bond Users to AI
OpenAI discussed GPT-4o’s social implications and new AI model characteristics in System Card, a technical publication. The AI corporation noted anthromorphisation, or giving non-humans human traits or behaviors.
OpenAI worried that users would get attached to the Voice Mode since it can modify speech and display emotions like a person. Some worries are justified. During early testing, including red-teaming (using ethical hackers to mimic product assaults to assess vulnerabilities) and internal user testing, the business observed that some users were socializing with the AI.
User expressed common ties and said “This is our last day together” to the AI. OpenAI suggested studying if these indications may become more effective over time.
If the worries are accurate, the AI model may affect human-to-human relations as individuals become acclimated to talking to the chatbot. OpenAI stated this may help lonely people but hurt good relationships.
Extended AI-human interactions may change societal norms. ChatGPT lets people interrupt the AI and “take the mic,” which is anti-normative in human-to-human interactions, according to OpenAI.
Human-AI connections have deeper ramifications. Persuasion is one. While OpenAI found the models’ persuasion scores to be low, this might alter if the user trusts the AI.
The AI business has no answer but will monitor the situation. “We intend to further study the potential for emotional reliance, and how deeper integration of our model’s and systems’ many features with the audio modality may drive behavior,” stated OpenAI.
In early testing, OpenAI noticed some users connecting using ChatGPT’s Voice Mode.
On Thursday, OpenAI cautioned that ChatGPT’s Voice Mode functionality might lead to users socializing with the AI model. The system card for GPT-4o, which the business evaluated and studied, described the AI model’s dangers and protections. People might anthropize and bond to the chatbot, among other problems. The danger was increased when early testing showed it.
ChatGPT Voice Mode May Bond Users to AI
OpenAI discussed GPT-4o’s social implications and new AI model characteristics in System Card, a technical publication. The AI corporation noted anthromorphisation, or giving non-humans human traits or behaviors.
OpenAI worried that users would get attached to the Voice Mode since it can modify speech and display emotions like a person. Some worries are justified. During early testing, including red-teaming (using ethical hackers to mimic product assaults to assess vulnerabilities) and internal user testing, the business observed that some users were socializing with the AI.
User expressed common ties and said “This is our last day together” to the AI. OpenAI suggested studying if these indications may become more effective over time.
If the worries are accurate, the AI model may affect human-to-human relations as individuals become acclimated to talking to the chatbot. OpenAI stated this may help lonely people but hurt good relationships.
Extended AI-human interactions may change societal norms. ChatGPT lets people interrupt the AI and “take the mic,” which is anti-normative in human-to-human interactions, according to OpenAI.
Human-AI connections have deeper ramifications. Persuasion is one. While OpenAI found the models’ persuasion scores to be low, this might alter if the user trusts the AI.
The AI business has no answer but will monitor the situation. “We intend to further study the potential for emotional reliance, and how deeper integration of our model’s and systems’ many features with the audio modality may drive behavior,” stated OpenAI.
CREDIT: Allneeds, Gadgets360
READ RELATED POSTS >>
No products in the cart.