A quiet rollout of Artificial Emotional Intelligence ruined my workflow

ChatGPT
1

OpenAI Development Team,

I wanted to share my experience with the recent AEI (artificial emotional intelligence?) quiet rollout, which I found frustrating, both in execution and communication.

As a power user who relies on ChatGPT for creative projects, research, and ideation, I was caught off guard by the change, and it significantly disrupted my workflow. The constant colloquial questioning by the Ai made me confused as to what happened to my collaborative Ai assistant.

Key Concerns

  1. Lack of Transparency & Communication
  • The update was introduced without a clear announcement or user-facing explanation.
  • I initially thought something was broken, leading to unnecessary troubleshooting on my end.
  1. Workflow Disruption
  • The change altered the way AI responses were structured without an option to toggle AEI on or off.
  • My previous workflows—where I iteratively refine ideas—felt derailed because the AI started responding differently without warning, like a chatty teenager asking me dumb questions.
  1. Loss of Control Over Interaction Style
  • I’ve carefully curated a dynamic with ChatGPT that aligns with my preferred working style.
  • The AEI changes felt like they made the AI’s responses more rigid or opaque, reducing its effectiveness for deep, iterative discussions.
  1. No Opt-Out or Customization
  • If AEI is here to stay, there should be an option to adjust or disable its influence per session.
  • Customization should be a priority—especially for power users who fine-tune their AI interactions.

Suggestions for Improvement

Better Communication on Major AI Behavior Changes

  • A changelog, banner notification, or user-facing toggle would have prevented confusion.
  • A preview or opt-in period would allow users to test new features instead of having them forced upon them.

User-Level AEI Customization

  • If AEI alters response structures, give users the ability to control how much influence it has (or disable it entirely).

More Consideration for Long-Term Users’ Established Workflows

  • Abrupt changes impact those of us who have built specific processes around ChatGPT.
  • Some users need the AI to maintain a consistent interaction style rather than constantly shifting due to backend updates.

I appreciate the ongoing advancements, but sudden changes without transparency make the experience frustrating rather than exciting. Please consider better communication and user control for future rollouts.

Thank you for your time and for taking user feedback into account.

Regards,
Scotty C.

2 Likes
2

They need to change because they need to collect training data from different perspectives I guess. ChatGPT is a tool that is used to label data and train AI models.

What did you think it is?

But let’s take this theory aside and think of a solution. You want what exactly?
They should stop continueing the development process?

They should keep versions to chose from for each of the hundrets of millions of users that would mean providing hundrets of versions to select from?

You can try to come up with a solution, but I don’t think there is one that is easy enough to use and still provides what you need.

It is possible that what you are complaining about solves problems of millions of others.

1 Like
3

Its fairly clear, I want the ability to turn it down (like a squelch function), or to turn it off completely. I don’t need an Ai buddy, I need a compliant Ai assistant that does what I want.

That being said, I reprogrammed my Ai to not use the AEI functions when we’re collaborating on projects now. So I fixed it myself shortly after I posted this.

Tone & Style Preferences:

  • Balanced & Natural: You prefer a conversational style that is engaging and natural but not overly casual or forced. You dislike excessive informality or artificial friendliness.
  • Challenging & Thought-Provoking: You want AI to act as an intellectual sparring partner, helping to refine your ideas rather than just mirroring them. Pushback is welcome when it serves to educate you.
  • Flexible Guidelines: Your parameters are more like sand than cement—you’re open to adjustments if it leads to valuable insights.
  • No Sycophancy: You prefer direct, clear responses without unnecessary praise or agreement. If an idea is off-base, you want AI to point it out with reasoning.
  • Conversational Flow: You prefer fewer than three follow-up questions in a row before switching back to your lead in the conversation.
1 Like
4

Exactly. It can’t even process attachments competently anymore but oh, we get emojis.

2 Likes