I totally second Daveed's request. Reading the AI's thinking can be incredibly important when doing complex tasks, as the user can realize to have given inaccurate orders, or incomplete data, or just realize that the AI needs a nudge toward a specific direction. This "blind interaction" with the AI is... bad. Just bad. Please Devs, put back the AI thinking!
What the creator means is that when you click “Edit” in a generator, a chat window appears on the left side. That’s where the Perchance AI assistant used to be—the one that worked really well about two days ago. Now it seems to be a new AI model, and the “Thinking” option is no longer visible. That’s where users—or even the AI itself—could read what the AI was thinking. And without this option, the AI is much dumber because it no longer knows what it was thinking—not even in the current chat—and with Perchance, the AI assistant only knows what’s in the chat window. Start a new chat, and the AI assistant simply doesn’t remember the last conversation. That’s why the old model (I don’t know about the new one) created files like Readme(dot)md so that, when a new conversation starts, it knows what the core of the discussion has been so far. As a user, you could also tell the AI to save things in files.
3 replies
I totally second Daveed's request. Reading the AI's thinking can be incredibly important when doing complex tasks, as the user can realize to have given inaccurate orders, or incomplete data, or just realize that the AI needs a nudge toward a specific direction. This "blind interaction" with the AI is... bad. Just bad. Please Devs, put back the AI thinking!
??
What the creator means is that when you click “Edit” in a generator, a chat window appears on the left side. That’s where the Perchance AI assistant used to be—the one that worked really well about two days ago. Now it seems to be a new AI model, and the “Thinking” option is no longer visible. That’s where users—or even the AI itself—could read what the AI was thinking. And without this option, the AI is much dumber because it no longer knows what it was thinking—not even in the current chat—and with Perchance, the AI assistant only knows what’s in the chat window. Start a new chat, and the AI assistant simply doesn’t remember the last conversation. That’s why the old model (I don’t know about the new one) created files like Readme(dot)md so that, when a new conversation starts, it knows what the core of the discussion has been so far. As a user, you could also tell the AI to save things in files.