The Future of AI: Is There a Place for Ambiguity?
As technologists and artists in new media (that strange place where words, imagery, meaning and new tech collide), we design, build, and test software experiences and products that center the deep creative impulse of humanity, because we are interested in the power of that impulse.
We’ve observed throughout our work that many computing and AI software and tools have underestimated, and even excluded, this human impulse from their design.
The particular joy of uncovering patterns and logic embedded in various functions of the universe, and building things that make use of them for people to enjoy, is unparalleled. It’s no surprise that as computing devices became more accessible and explainable, people fell in love with them. But today, many are becoming uncomfortable with their ubiquity and feel less understood and more constrained by computing and AI, and that they must fight or compete with the software and products sold to them.
Computing and AI design has been commandeered by business needs and goals, which seek to predict a person’s actions — will they buy x product or y product? Through a business lens, the ultimate goal is to crack and tame the unpredictable nature of people, laying product after product in front of them, so enticed, they scoop them up one after the other, like Hansel and Gretel did the witch’s candies.
People don’t appreciate being hoodwinked, but they do appreciate ease, and companies have become excellent at hiding the latter with the former — Alexa can compile shopping lists, answer questions and respond to commands and learn from a user’s inputs, which makes life ‘easier’ but Amazon will also use that learning to sell data — a composite of a user’s movements, purchases, desires, and beliefs — on to other companies, and to (frighteningly) governments.
No one wants to be reduced to a dollar sign, or be made a tool merely for more profits. We need agency and freedom, and yet computing and artificial intelligence design has been laser focused on creating products and software that meet business needs and goals, while leading us down the garden path to do it.
Computing and AI have historically been designed for precision. Early computers were incubated in research departments funded by government militaries, and those designs had applications in space exploration, medical research and agriculture, all areas not explicitly funded by military departments, but that relied on military designs as a starting point, which heavily favored precision.
For example, during WWII, ground soldiers routinely missed aircraft targets, so military manufacturers integrated software into newly issued weapons to increase the soldiers’ target precision. Suddenly they began to accurately hit aircraft targets. One wonders if they knew the real reason why.
Some may call this ability to increase precision valuable, but after WWII, when software-assisted decision-making, originally designed for war, entered the realm of cultural production, the measure of its value became questionable. If we extrapolate this history and survey where computing and AI has taken us, it’s clear that we’ve relied heavily on the design element that produces precision — the loop.
