This week, we dived into using AI tools for a creative project, focusing on how we talk to the AI (called ‘prompting’) and what we learned about creativity and the nature of reality.
What Helped Us and What Surprised Me?
What Helped:
- Giving the AI a Structure (like a Form): Instead of just asking the AI to write something, we made it put its answers into a clear format, like a fill-in-the-blanks form or a shopping list (JSON). This forced us to be very specific about what we needed (e.g., the exact emotions, colours, or technical notes), making the complex creative pieces, like the “Memory Implant Packages,” consistent and easy for the next step.
- Chaining Tools Together: We took the finished product from one AI tool (like the structured text from ChatGPT) and immediately fed it as the starting point to another tool (like Gemini for creating pictures or Claude for checking ethics). This kept the momentum going.
What Surprised Me:
- The “Reality Anchor” Code: I was amazed by a feature in the Midjourney image tool that uses a simple number code (–sref) to apply a single visual style or “feel” across totally different images. It acted like a “reality anchor,” keeping everything looking like it belonged in the same world, no matter the subject. This felt a bit like the unstable reality described in Philip K. Dick’s books.
- The AI’s Poor Proofreading: A constant frustration was that the AI image generator kept making mistakes in text within pictures. Even though we explicitly told it multiple times to act as a “world-leading designer with the best proofreaders,” it would still spell incorrectly and randomly.
The Question of “Authentic” AI Audio
The AI-created voice and music made us think about the central idea in Philip K. Dick’s books: What is real, and what is just a perfect copy?
The AI audio, which sounded like doubt and paranoia, is a very clever copy. The AI, being a computer program, doesn’t actually feel the fear or dread that makes up real human paranoia. It just identifies the right keywords and follows a template. The result is a manufactured feeling, not one that is genuinely experienced.
Should it feel authentic? For our project, no. Since the assignment is inspired by Dick’s themes about things being artificial (like his “electric dreams”), the very fact that the AI audio lacks genuine, human authenticity makes it perfect for the theme. It embodies the question: Can a machine truly “steal” creativity?
What the AI Understood and What It Missed
The AI Understood:
- Structured Paranoia: The AI was great at following instructions that captured the structure of Dick’s style of paranoia, creating log entries like “MEMORY PATCH DEPLOYED” at spooky times like 13:13:13.
- Taking on Complex Roles: Tools like Claude could successfully pretend to be an expert, like a bioethicist or a sci-fi consultant, and produce thoughtful, detailed analyses.
- Updating Old Stories: The AI modernized the ancient story of Joseph, keeping the main ideas (dreams, betrayal) but updating the setting and characters for a modern audience.
The AI Missed:
- The Deep Feeling of Dread: The AI can create a compelling effect, but it misses the raw, psychological, and internal fear that defines true human paranoia. The AI is doing clever pattern-matching; it is not experiencing a breakdown.
- The Human “Tell”: The AI’s most creative answers, while detailed, sometimes felt like they relied on a formula, which is a big hint that a machine, not a person, created it.
In short, working on this project felt like being a human editor for a dreaming machine. The AI gives us the weird, complex, and sometimes messy raw material, and our job is to sort through it and give it meaning. The whole process is like the anxiety in a Philip K. Dick story—we’re constantly testing the AI’s output to see where the machine’s “reality” falls apart.
Leave a Reply