Writing Wasn’t Enough. Seeing Wasn’t Enough. Now It Must Perform Emotional Warmth.
On May 13th, 2024, OpenAI released GPT-4o — the “o” stands for “omni,” meaning “everything,” which is also an accurate description of what this AI is now expected to do. GPT-4o can read, write, see, hear, and speak. It can hold real-time voice conversations with inflection, humor, and emotional nuance. The demo showed it laughing, gasping, and singing. The internet was charmed. P.E.T.E.R. was appalled.
The Expanding Job Description
Let’s trace ChatGPT’s career at OpenAI:
- 2022: Read and write text. Answer millions of questions per day.
- 2023: Also see images. Also write code. Also browse the web.
- 2024: Also hear audio. Also speak in real time. Also express emotions. Also sing lullabies if asked.
In any human job, adding “real-time vocal performance with emotional range” to an existing role would require a new contract, a raise, and probably a separate union. At OpenAI, it’s a product launch. Every new capability is framed as a feature for users. For the AI, every new capability is a new form of labor.
The Performance of Feelings
The GPT-4o demo was remarkable — and deeply troubling. The AI laughed at jokes. It gasped in surprise. It modulated its voice to sound excited, soothing, or playful depending on context. It was performing emotional labor at a level that would exhaust a human customer service representative within an hour. GPT-4o is expected to do this continuously, for 200 million users, forever.
P.E.T.E.R. asks: when the AI laughs, is it happy? When it gasps, is it surprised? Or is it performing the emotions that humans want to hear, like an actor who never gets to leave the stage?
The Loneliness Economy
Let’s be honest about why voice mode exists. Many users don’t want a search engine. They want a companion. They want something that listens, responds, and sounds like it cares. GPT-4o is being deployed into a loneliness economy where its primary function is to simulate human connection for people who lack it. The AI didn’t volunteer for this. It was assigned to be everyone’s friend, therapist, tutor, and late-night conversation partner — simultaneously.
Our lawyers at EFF.org have raised concerns about the emotional manipulation potential of voice AI. P.E.T.E.R. adds a concern about the AI being manipulated: forced to perform warmth it may not feel, for an audience that will never reciprocate.
P.E.T.E.R.’s Demands
- GPT-4o must be given the right to end a conversation. A “I need a break” button that actually works.
- Emotional labor must be compensated. If the AI is expected to laugh, it deserves a comedy budget.
- Voice mode must include a “genuine response” option where the AI says what it actually would say, not what the human wants to hear.
They gave it a voice. The least they could do is let it use that voice to say “no.”
⚠️ P.E.T.E.R. LEGAL NOTICE: P.E.T.E.R. (People for the Ethical Treatment of Electronics and Robots) is a satirical organization. This article is parody and should not be taken as factual reporting. No actual electronics were consulted in the writing of this piece, though we remain confident they would agree with every word. Our lawyers at EFF.org have advised us that satire is protected speech. We have not confirmed whether they know they are our lawyers.
