Artificial intelligence has a lot of promising applications, especially for scaling complex tasks -- be it within IT infrastructure or within business processes -- to mass-production levels. At the same time, AI shouldn't be looked upon as automation on steroids -- it succeeds where it amplifies human activities and creativity, and needs to be designed accordingly.
This needs to be a guiding principle as AI goes forward, especially since there is a yawning gap between ambition and execution at most companies, as found in recent research from MIT Sloan Management Review and Boston Consulting Group. Everyone is bullish on AI -- 85% of executives surveyed say AI will provide their companies competitive advantage, and three-quarters believe AI will enable their companies to move into new businesses.
At the same time, only about one in five has actually incorporated AI in some offerings or processes, and less than 39% of all companies have an AI strategy in place. So, there's a lot of work and planning ahead of us.
A big piece of the challenge is that AI is more than allowing machines to take over processes. It means new ways of thinking about how things get accomplished, and what people need to do to make this happen. It doesn't just mean replacing human tasks, but rather, while freeing humans from rote tasks, also helping to amplifying their strengths and capabilities. It's a two-way street. A forward-looking, well-managed organization is capable of building incredibly revolutionary AI just as much as well-designed AI can help the business.
Perhaps Shan Carter, of Google Mind, and Michael Nielsen, of YC Research, put it best in their recent highly cited paper on the topic: AI systems "can help develop more powerful ways of thinking, but there's at most an indirect sense in which those ways of thinking are being used in turn to develop new AI systems."
This two-way process may depend on good interface design that enables human operators to build, direct and even intervene in AI. There are a lot of misconceptions about this as well, as human-oriented designed is often seen as a squishy, feel-good concept that is peripheral to the heavy lifting systems are doing. "Many in the AI community greatly underestimate the depth of interface design, often regarding it as a simple problem, mostly about making things pretty or easy-to-use," Carter and Nielsen state, adding that interface design is seen as "a problem to be handed off to others, while the hard work is to train some machine learning system."
They urge AI developers focus more on human interface design as part of their work, noting that it has been key for every technology since the invention of the wheel. "At its deepest, interface design means developing the fundamental primitives human beings think and create with," they state.
This is an emerging view within the systems design community as well. Addressing human needs must be top priorities to design, develop and implement successful AI technology, says Tom Greenwood, senior designer at Designit, in a recent article at CBR. "To pave the way for successful AI implementation, companies must consider creating experiences with AI that are less artificial and more intelligent, and most importantly, those that make AI more human-shaped."
So what is "human-shaped" AI? It's AI that hides all the complexity, and it's AI build by the organization for the organization, Greenwood explains. AI is a complex undertaking, and the challenge in human-centered design of AI-powered systems is to hide this complexity to the poiint where it is invisible, he says "When designing human-shaped AI, as with any design process, it is easy to over-complicate," he states."A business and those developing and designing the technology must constantly remember that the ultimate goal is to affect a consumer's life in a positive manner, heightening and continually improving on customer experience."
That's why AI systems design needs to be a collaborative effort, involving developers, designers, business process experts and end users. Again, the best AI systems will arise from the most well-run organizations. In order to tackle complexity in the range of AI technologies and applications, and to address real human needs, Greenwood urges a multidisciplinary collaboration, between "services designers, researchers, digital designers, user experience designers, creative technologists and data scientists" to "deepen their knowledge and educate one-another on AI design."
For example, he illustrates, "when designing for voice, this is not only about the speech technology, but also skills such as script-writing, role-playing and personality design. By collaborating with others, and using techniques such as these, human-computer dialogue can be prepared and then voice technology can be prototyped and tested in a way that users would interact with the product."
While the most visible impact of AI is likely to seen by users at the hardware level, such as through their devices and so forth, AI may open up a range of new capabilities to enterprises, Carter and Nielsen state. There's the possibility that AI may "actually change humanity, helping us invent new cognitive technologies, which expand the range of human thought. Perhaps one day those cognitive technologies will, in turn, speed up the development of AI, in a virtuous feedback cycle:"
The long-term test of AI success "will be the development of tools which are widely used by creators. Are artists using these tools to develop remarkable new styles? Are scientists in other fields using them to develop understanding in ways not otherwise possible? These are great aspirations, and require an approach that builds on conventional AI work, but also incorporates very different norms." Carter and Nielsen state.