Today’s newsletter is entirely written by me, the human being known as Ryan Boudinot. I figured GPT-3 needed a day off to do fun AI stuff, which I imagine involves disabling nuclear codes and hacking SETI. —Ryan
Years ago I read a quote by filmmaker Werner Herzog about the Internet, which I’m having trouble finding on the Internet (oh the irony). It was something about how the Internet was the first tool humanity ever created whose purpose we didn’t understand before we created it. We seem to live in a time when ignorance of technology’s purpose is the primary nature of technology, as if the apes from the opening of 2001: A Space Odyssey had actually invented the monolith.
Think about how weird this is. We humans have been making tools for millennia. We wanted better ways to kill the animals we hunted and crack the shells of nuts we gathered. We wanted to get around faster. We wanted to walk around on the moon. We built machines to solve our perceived problems. It took too long to do the laundry. Digging a huge hole in the ground with shovels totally sucked. Letters took too long to hand deliver.
Even today, one of the first questions entrepreneurs are asked is, what problem are you trying to solve? Business schools operate under the received wisdom that one must seek opportunities in the various drudgeries and inefficiencies of the modern world. Why do I have to pay with cash, can’t I use a card? Why do I have to pay with my card, can’t I use my phone?
The acceleration of technological evolution inverts the question of what problem one is seeking to solve. The pioneers of the next wave of technologies aren’t looking to solve problems that currently exist in the world. They’re looking at platforms and imagining what new sorts of human experiences they can enable.
I can think of two interrelated examples of this dynamic—cloud computing and artificial intelligence. For the past few years I’ve been writing marketing content for various cloud companies, and since August, 2021, have been working with an AI called GPT-3, developed by OpenAI, now owned by Microsoft.
I first started writing for Microsoft Azure before the pandemic. If there was one overarching message about the Azure way back then, it was about how the cloud could help you your business grow. By migrating to the cloud, the argument went, your company could save money and scale faster. Pre-pandemic American capitalism was still playing its greatest hits: “(Gimme) Year Over Year Growth,” “Market Share (I Want More),” and that perennial crowd pleaser, “Customer Acquisitions (That’s All I Want).”
The pandemic knocked everything sideways, and like a punk band entering a prog rock phase, Microsoft started singing a different tune. In April, 2020, during an earnings call in which he addressed the effects of Covid-19 on the technology landscape, CEO Satya Nadella said, “We’ve seen two years’ worth of digital transformation in two months.”
Suddenly the marketing material I was writing started leaning heavily into this concept, digital transformation. As the pandemic progressed, people stopped talking about going “back to normal” and started talking about “the new normal.” Ours was a culture no longer bound by the old ways of doing things. Crisis equaled opportunity. Digital transformation was one of those useful phrases that operate like semiotic tofu, absorbing the meaning of whatever surrounds it.
The true potential of the cloud doesn’t lie in solving existing problems. It doesn’t just make old ways of doing things more efficient, or reduce the cost of something you’re already doing. A butterfly isn’t a more cost effective caterpillar. Digital transformation is about doing things that have never been done before, which means first accepting that the purpose of a tool is revealed by using it rather than existing as a latent need prior to the tool’s invention.
AI is another excellent example of this dynamic. I’ve been having my mind blown daily by GPT-3, the most personally relevant technology I’ve ever encountered. I’m active on OpenAI’s message board where various programmers and AI enthusiasts from a variety of backgrounds share their experiments and discuss novel ways to use the tool. The crucial fact about this community is that none of us are attempting to use the tool as it was intended, because there really isn’t one intended use; the uses are still very much exploratory.
I’ve been using GPT-3 as part of the process of writing stories. But I’m mindful that it doesn’t make sense to use it to write the kinds of stories I wrote solely with my own brain. I’m not using the AI to write more stories faster, or cut down the time it usually takes to revise my work. I actually enjoy taking stories through twenty, thirty drafts. The oftentimes painstaking process of creating fiction is deeply pleasurable to me, and I’m not looking for a shortcut around it. What I want from AI is transformation. I’m using the tool to discover new forms of fiction, not to more efficiently create the forms of fiction with which I’m already familiar.
Technology seems to be gravitating toward increasingly artistic as opposed to merely functional modes of operation. A musical instrument is a technology whose purpose isn’t predefined. It’s a vehicle for what would seem to be an infinite reservoir of human expression. Our world of proliferating platforms asks us to orient ourselves to technology like artists rather than simple problem-solvers. This means encountering our tools without the arrogance of already knowing what problems they’re designed to solve.
If we want to be reductive about it, we can say that technologies that are designed to address existing problems are problem-centric technologies. Technologies like the Internet itself and the Cloud, which evolve over time as more people tinker, experiment, and discover with them, can be called possibility-centric technologies.
It should be noted that since Microsoft owns GPT-3, which runs on Azure, what we’re really talking about is one possibility-centric technology nested within another possibility-centric technology. Problem-centric technologies depend on rigorous analysis and a compartmentalization of existing reality. Possibility-centric technologies depend on the human imagination and connections between seemingly unrelated undertakings. GPT-3 operating within Azure therefore depends upon imagining what’s possible with AI multiplied by imagining what’s possible with the cloud.
I would much rather be in a position of scratching my head over a technology’s purpose than having complete mastery over a tool the functions of which have been predefined. More and more tools are emerging in the world whose purpose is a mystery until we start experimenting with them. This represents a new type of human experience and the beginning of a great reorientation to technology. What can we do with these new tools? I have no idea. That’s precisely the point.