How can we prompt biased AI assistants to help us think different and imagine diverse tech futures?

In this episode, we invite all the crazy ones, the misfits, the rebels, the troublemakers to think different with us, like Apple used to advertise in the late 90ies. But instead of worshiping intelligence and individual genius, we want to dance with the stochastic imagination of generative AI tools. We wonder how we might collectively go beyond prompt engineering that’s focused on productivity and getting answers –  fast! –  to prompting diverse ways of knowing and being, while remaining mindful of current AI biases and limitations. 

We reflect on our experiments with woo prompting and custom instructions, and share our observations on the power of conversational learning and language. Our experiments show that being an active participant in conversations with AI chatbots can yield more diverse outputs, but it’s important to keep steering outputs from the default, most statistically probable, to more diverse perspectives. We see role-playing and developing questioning skills as useful strategies for getting more diverse outputs. 

We also reflect on what we learned from having chatbots chat with each other, what makes ChatGPT computationally tickled, and why we think AI assistants should get better at asking thought-provoking questions, rather than rushing to answers and people pleasing. We ponder on the time component of our conversations and how both humans and machines might benefit from having more time to engage in slower, systems 2 thinking to collaboratively explore our biases and assumptions instead of jumping to quick solutions.   

If you’d like to wonder and wander with us, join Tethix firekeepers Alja Isaković and Mathew Mytka in this meandering exploration inspired by the latest Pathfinders Newmoonsletter. In addition to the resources explored in the Newmoonsletter, we recommend exploring:

• Mat’s LinkedIn post on experimenting with custom instructions

• Alja’s LinkedIn post on getting ChatGPT to imagine an alternate tech future 

• Are we ready to navigate the complex ethics of advanced AI assistants? by Andrew Maynard

• Should We Respect LLMs? A Cross-Lingual Study on the Influence of Prompt Politeness on LLM Performance

• Questions: Find your voice with mmhmm’s AI-enhanced interviewing tool

• Do LLMs Have Distinct and Consistent Personality? TRAIT: Personality Testset designed for LLMs with Psychometrics

• Taxonomy of Tethix Mirrors and our  IEEE article on Rainbow Mirrors

• System 1 and System 2 Thinking

• Vanessa Andreotti: "Hospicing Modernity and Rehabilitating Humanity" | The Great Simplification 125

• Experimental Rainbow Mirrors GPT

You can learn more about the Tethix pathfinding adventure at: https://tethix.co/pathfinders/