Designing in an age of mechanical intelligence
New technologies have the potential to reveal the deep strategic value of design, says Jason Walsh
Artificial intelligence (AI) has consumed acres of newsprint in recent months. Everyone knows it will have an impact on business, but precisely what that will be has been less clear. Naturally, this has created a boom time for commentators, but it can also make considering its potential impact a difficult task.
Chris Donnelly, user experience principal at design agency Each&Other, said that his interest had been piqued by the technology and the possibilities it could create. “We’re watching the space and playing with large language model AI and learning about it.”
Most recent discussion of the new developments in AI falls into one of two camps: fear of its job-destroying potential on the one hand and, on the other, its potential to radically speed all manner of processes, including application development.
The truth is, though, no-one really knows how it will develop or what effects it will have on society. As a result, Donnelly said, rather than responding with panic or over-excitement, now is the time to ask fundamental questions about how machine intelligence and automation could augment what we do.
“If this is the direction of travel, then it will have significant consequences for business. What's in my mind at the minute is: what is the role of the design team?” he said.
Donnelly said that Each&Other’s mission is likely to be augmented by the adoption of AI.
“You still need to understand the problem you are trying to solve. You have to understand what quality is and what good looks like. We’re working with companies that tend to put out software products, and a bad product will fail,” he said.
Indeed, designers in general, regardless of discipline, have long been misunderstood. Although they all receive extensive visual education, there is a lot more to design than drawing. As a result, AI has the potential to make clear the strategic value of design, not just in prototyping or execution, but also in deep research.
“Historically, the thing that has been harder to sell is the research element. If we commoditise parts of design [process] then the most important piece will be getting inside users' heads. It will become more strategic,” he said.
How AI will develop in the coming years, and indeed months, will answer whether or not this transformation happens.
“Will it be a niche, interesting technology, like crypto? Will it be like the iPhone and change how businesses do business and how we develop apps? Or will it change the course of history like the invention of the printing press?” said Donnelly.
While there are no guarantees, thus far, progress has been rapid, with claims that the latest iteration of GPT, version 4, is ten times more powerful than its predecessor, version 3.5 (better known as ChatGPT).
“That's a bit like going from a Commodore 64 to an iPhone in six months,” Donnelly said.
Open AI’s ChatGPT is only the best-known of the current crop of AIs. Others, including Google Bard and a recent demonstration from Adobe, are also set to challenge how we work. Donnelly said that their true value will be revealed not by unleashing them on the open web so much as using them to make sense of the giant lakes of data that businesses float on.
“The consensus is [that] there's going to be a couple of very large LLMs, GPT and a couple of others, and you'll build a middle layer on top of that, and that will be your data which it will read and interpret.”
The significance of this is that much data is currently siloed and unstructured, making it hard to extract value from. An AI, however, would be capable of sorting through it.
“That would make your data your differentiated product,” Donnelly said.
There are limitations to today’s AI, however, including one that should worry anyone seeking to use it in a business context: hallucination. In other words, it lies, including occasionally making up fake references to back up its presentation of false facts.
“The other thing that's interesting is, GPT makes things up. It’s a very strange phenomenon,” Donnelly said.
Nevertheless, Donnelly said, while the technology has a way to go, it has already demonstrated very significant potential.
“I think the character of work is going to change,” he said.