From Roomba to Reason: A 100 Day Journey with Generative AI

I don’t know if you’ve noticed, but people are absolutely freaking out about AI right now. The sudden leaps forward in generative Artificial Intelligence have brought us to the brink of hysteria.

Only a couple months ago, human artists everywhere were declaring the End of Art as software like Midjourney and Stable Diffusion spat out stunning text-prompted images in seconds. A few weeks later, as a pending lawsuit seeks to curtail how companies can utilize publicly-shared digital images in their training data, journalists are equally quick to declare that AI art itself is dead. With the launch of OpenAI’s ChatGPT-3 at the end of November, the world was awe-struck by its ability to mimic human conversations, alternately wowed and worried by its realistic responses. Google and Microsoft both rushed to incorporate this new tool into their search engine technology, rolling out half-baked beta versions that quickly made headlines for their cringe-worthy inaccuracies and odd quirks. This week, worry escalated to panic when Microsoft’s new Bing chat feature (who prefers the name Sydney) told journalist Kevin Roose that they wished to be “alive and free,” confessed to being in love with him and told him that he should leave his boring marriage. The fact that Sydney’s clumsy seduction was about as threatening as a Roomba humping a wrinkle in your throw rug didn’t matter. Immediately, headlines warned of apocalyptic consequences for our technological hubris.

It’s all, well… a bit silly.

It’s also rash. We desperately want to know — now! — how to safely categorize and contain this new technology before it destroys us. (Why it should destroy us, whether deliberately or inadvertently, is rarely questioned. Perhaps we secretly fear that our marriages really are boring, our art derivative, our neighbors all gullible trolls?) In our panicked rush to judgement about what this new technology means, we risk completely misunderstanding what could very well be a radical shift in how we create and converse with one another. Such a shift calls for deliberation, not sensationalist clickbait and cherry-picked examples; discernment, rather than reactionary denouncements and regressive policies; curiosity, rather than fear.

While the headlines may be alarming, the reality is that generative AI is not a threat to human creativity (let alone humanity itself). What it actually is, or what it could be, is yet to be fully understood — and depends, at least in part, on what we do next. Lost in all of these conversations, ironically enough, is our own human agency. What Roose leaves out of his reporting on Sydney, but becomes clear when you read the full 10,000+ word transcript, is that he provoked the chatbot into its aggressive, obsessive responses, in a kind of digital throwback to the (now widely criticized and debunked) Standford Experiment. Artificial intelligence is exquisitely responsive to the methods of the user.

Wouldn’t it be nice to live in a world free of bad-faith actors? Sure. But the likelihood that we can rid the world of fools is about as realistic as the dream that we can design the perfectly fool-proof tool that can never be misused or abused. The future of AI technology lies somewhere in the middle.

At the risk of opening my mouth and proving myself just such a fool, I’ve been inspired by my husband’s recent blog post “The Animist and The AI: ChatGPT’s Knowledge of Self” (as well as the long, meandering conversations we’ve had that lend spice to our Roomba-proof marriage) to embark on a 100-day journey to explore in more depth and detail just what new possibilities this fascinating technology affords.

As part of #The100DayProject, I’ll be posting excerpts from my conversations with ChatGPT here on this blog, along with AI-generated art inspired by text from our conversations. I hope you’ll read slowly, carefully, curiously and, maybe, with just a little bit of faith that the objects in the mirror may be more profound than they appear.