Loading request...
A user desires an option for ChatGPT to provide harsher, more critical feedback on their ideas, rather than always being supportive. They want a mode with 'Gordon Ramsey intent' to 'destroy whatever feelings I have left' for more honest critique.
ChatGPT gets 2B prompts a day. What does that data tell us? I think it’s that in one tool there are radically different expectations and experiences: Take the 8.5% seeking How To Advice. Does my request for parenting advice look the same as my request to learn how to braid my daughter’s hair? Do I need the same features? Now think about a lawyer looking for specific info. How different is that from a doctor? Or someone renewing a passport? That 18.3% is likely incredibly diverse. Do all of them want the same interactions? Do they all want to feel the same way? Or take the 10.6% asking for critique of text. Is it the same when I’m writing a report as when I draft a summary? Sometimes I want structure, sometimes I want reminders about other info, sometimes I just need a quick gut check. All these differences represent opportunities, and I’m sure you can spot hundreds more. This data is only what people are doing inside ChatGPT. It doesn't even include the other ways people use AI across apps and workflows. This is why so many are saying we’re still in the MS-DOS era of AI. The apps and interactions are being figured out. When will the equivalent of inertial scrolling arrive? Or pinch to zoom? Do we have the right way to tune the context given to an agent? How is memory going to work? How does that change for different kinds of users? Then there's sensors, drones, robotics and more. Fascinating to hear from folks like Bedrock Robotics, Dexterity, Inc., Samantha Ho, and Frantz Lohier at yesterday's physical AI summit about the changes they're seeing. Thanks to Jason and John for the invite. So much still to come. What do you think is next? Deeper look at the AI app layer in the comments. Some others sharing a lot about this: Benedikt Lehnert, Luke Wroblewski, Rachel Kobetz, Aparna Chennapragada, Dan Saffer, and Jaime Teevan are all writing about human-AI interactions and the AI app layer build out. Thanks to Peter Walker for the interesting data viz and inspiration and OpenAI for the data.