PANIC!

Sign in

PANIC!: Playground Ai Network for Interactive Creativity

PANIC! is an interactive installation where you can play with feedback loops of generative AI (GenAI) models hooked up end-to-end. As well as generating intriguing text, images and audio, PANIC! explores how different ways of connecting these models up can give rise to different patterns of outputs, emergent behaviours, recurring patterns, and degenerate cases. Today, the low barrier to entry in GenAI model platforms (in terms of cost, time, and knowledge) means that more and more of us are using them. But just because we can put our text/images/audio into these models and have them provide new text/images/audio in return, does it mean that we should?

Instructions

Panic is an interactive installation, and we invite you to give it a try:

  1. type something (anything!) on the keyboard, or scan the QR code in the corner of the first screen and type it on your phone
  2. hit Enter & get ready to PANIC!
  3. keep watching to see the initial output, then the subsequent outputs
  4. what questions does this raise about feedback loops in GenAI-augmented systems?

Reading the screens

Your prompt goes to the first model in the network, its output goes to the second, and so on round the loop until someone types a new prompt. The panel beside the QR code shows the network's models in order (the one at work right now is highlighted), the initial prompt, how far the run has got and roughly how much energy it has used so far.

Each of the other screens shows one model's latest output, labelled with that model and the one whose output it was made from. When that input was text, the label quotes it, so you can read the exact prompt an image was made from and compare the two.

PANIC! Research Questions

At the School of Cybernetics we love thinking about the way that feedback loops (and the the connections between things) define the behaviour of the systems in which we live, work and create. That interest sits behind the design of PANIC! as a tool for making (and breaking!) networks of hosted generative AI models.

Anyone who's played with (or watched others play with) PANIC! has probably had one of these questions cross their mind at some point.

One goal in building PANIC is to provide answers to these questions which are both quantifiable and satisfying (i.e. it feels like they represent deeper truths about the process).

Maybe you've got some ideas about answers? Chat to the friendly folks at the PANIC! booth and let us know.

how did it get here from that initial prompt?

  • was it predictable that it would end up here?
  • how sensitive is it to the input, i.e. would it still have ended up here with a slightly different prompt?
  • how sensitive is it to the random seed(s) of the models?

is it stuck?

  • the text/images it's generating now seem to be "semantically stable"; will it ever move on to a different thing?
  • can we predict in advance which initial prompts lead to a "stuck" trajectory?

has it done this before?

  • how similar is this run's trajectory to previous runs?
  • what determines whether they'll be similar? the initial prompt, or something else?

which parts of the system have the biggest impact on what happens?

  • does a certain GenAI model "dominate" the behaviour of the network? or is the prompt more important? or the random seed? or is it an emergent property of the interactions between all models in the network?

How much energy does it use?

The screens show a running estimate of the energy each run has used so far, compared with everyday things like a 10 W LED bulb, charging a phone or boiling a kettle. None of the platforms report how much energy their models use, so the figures are ballpark estimates built from GPU time and published measurements, and they leave out the energy used to train the models. How we estimate it.

Techy stuff

Panic is the work of Dr. Ben Swift as part of his job at the ANU School of Cybernetics. The core "engine" is an (Ash + Phoenix LiveView) Elixir app, and it talks to various different hosted GenAI platforms (e.g. Replicate) to run GenAI models created by OpenAI, Anthropic, Meta, Amazon, Google, Microsoft and more. The outputs of these models (text, images and audio) are displayed live, in real-time, for viewers on screens anywhere in the world.

Contact

If you'd like to know more about PANIC!, contact Ben at ben.swift@anu.edu.au. If you'd like to keep up-to-date about the School of Cybernetics, you can sign up for the mailing list.