There was a time when the mere mention of artificial intelligence was wrapped in constant debate and triggered images of Hollywood-crafted products, like Hal 9000. The concept itself is quite controversial; it challenges human thought as Darwin once challenged human origins. But we moved on, and now we carry these intelligent machines in our pockets.
There’s a 38.9% chance you have one, too. Siri, the out-of-sight personal assistant from Apple, delivers an amazing experience. It listens to you, understands you, does what you say, and even talks back to you.
Sounds simple enough for us humans, but these are remarkable achievements for a machine. It has to process language, interpret context, understand intent, and orchestrate multiple services and information sources. And it brings together technologies that rely on dialog and natural language understandings, machine learning, evidential and probabilistic reasoning, ontology and knowledge representation, planning, and service delegation to do it.
Spin back the clock 50 years and all of this wasn’t even remotely possible. But just two years after Turing published the first documented idea of intelligent machines, three people were already working on the first system capable of speech recognition, named Audrey.
It could only process digits. Spoken by a single voice. With pauses in between. And it occupied a six-foot high relay rack.
Not exactly a marvel of technology, by today’ standards. But back then, when computers had only 1kb of RAM, it was an impressive achievement. More impressive still, when you think about how such a system came to be.
It all started with an illusion act
Many elements from very different spheres come together in the story of Siri, and it all starts with a man doing some magic.
Tracing Siri’s ancestry takes us back roughly 250 years, to Austria, when Vienna still had an empress. The story begins with a man known mostly for what was perhaps the most famous illusion in history: the Mechanical Turk, a machine that could play chess on its own and claimed to win over any opponent.
In reality, it was just a wooden cabinet with a life-size, mustache-wearing doll on top and a man inside, playing chess. It tricked people into thinking the machine was intelligent, but the idea itself was enough to intrigue the likes of Napoleon. (He played the Turk and he lost.)
And while the Turk made its creator—Wolfgang von Kempelen—popular, it is another of von Kempelen’s inventions that marks the beginning for Siri’s story.
The first speaking machine was a pretty straight-forward concept that tried to simulate the human vocal tracts—it had lungs and everything. Nevertheless, it was the first machine that could replicate whole words and sentences. It was this machine that would set the stage for Audrey.
Chess, the game that made it all possible
von Kempleton’s Turk was the first machine that could replicate human speech. Audrey was the first that could recognize human speech. But Siri is the first machine that can *understand human speech.
Understanding is the unique ability that swings the story back to the Turk. The machine’s connection with chess isn’t random. Chess is more than a game; it’s an entirely mental activity. And it’s a perfect metaphor that would allow for the birth of a new scientific discipline, artificial intelligence.
A machine capable of defeating a human opponent at a mind game is an intelligent machine, by any logical standards—or, at least, that was the premise.
While the Turk was, for the first time in history, the first real image of a machine that could be better than us at anything, it was just an illusion with a man operating it. But ever since, the idea of an intelligent machine started slowly morphing into physical technologies.
The next obvious stage would certainly seem to be a machine that could play chess and be self-operated. In 1912, the real thing quickly followed. It was called Ajedrecista and it was the first computer game. Only, without an actual, you know, computer.
Making this happen required a deep understanding of how we think when we play chess.
Every move weaves together an amazing chain of mental processes: Perception transforms the pieces on the board into a series of symbols, and long-term memory overlaps perceptions with previous knowledge. Logical thought then searches for variations, and decision-making is needed for the actual move.
Move after move, the chess game becomes a sequence of decision-making events governed by strict logical rules. And it is this logic module in our brain that chess heavily stimulates, so much so that it can be simulated. It doesn’t take a big imaginary leap to imagine that thought can be simulated.
This realization gave way to wonderful theoretical breakthroughs. Concepts like an algorithms, recursiveness and programming were born. Having to analyze how we think about chess quickly lead to computer thinking.
AI: A new, old way of designing experiences
A special group of people made a great imaginative leap. They realized that a game holds the secret into human thought. For people like Edward Feigenbaum, Marvin Minsky, Allen Newell, Herbert Simon, Alan Turing, John von Neumann, and Norbert Wiener—the founders of AI as a scientific discipline—pinpointing all the mental processes that are necessary to generate high-level cognitive activities played a very important role in the development of simulated thought processes through computer programming.
Logic and process alone wasn’t enough though. We expanded our concepts to expert systems, knowledge engineering, neural networks, and so on. The subsequent knowledge-based models of thought are nothing short of amazing. But the real breakthrough came from an anti-type of approach: The father of expert systems, Edward Feigenbaum, called it representation. This approach supported the idea that knowledge-modeling the real world was much too difficult; instead, systems should adapt and respond effectively to real interactions with the world.
This is important because it has finally allowed for the development of a truly human-centered approach to designing systems, an approach initially articulated by Bill Moggridge and one which inspired a major shift in design thinking that we see maturing today.
AI and HCI have been described as having opposite views on how humans and computers should interact. Human-centered computing is somewhat bringing all that together by combining intelligent systems, human-computer interaction, and contextual design. Instead of trying to imitate (or substitute) the human, the goal is to amplify and extend his capabilities, much like a prostheses does, although not in the sense that they compensate for the specific disabilities of any given individual, but rather because they enable us to overcome the biological limitations shared by all of us.
Above all else, a prostheses needs to fit, otherwise it will be rejected. In the same manner, systems designed to assist, rather than replace, need to be personal and contextual. They need to be intelligent in order to fit.
In terms of actual capabilities, Siri wouldn’t pass a Turing Test. But it doesn’t set out to do so. It doesn’t try to augment our abilities, but rather extend them.
For example, say you want to go to the best restaurant around. You know you can do that. With the help of technology, you can combine information from different sources (local business directories, geospatial databases, restaurant guides, restaurant review sources, online reservation services, and your own favorites).
But why would you want to? You want to use technology as a tool, not get immersed in the experience of interacting with it.
Siri delegates everything you don’t want to do. It lets you use technology as it’s supposed to be used, as a tool. By doing so, it becomes a digital prostheses. As a result, the experience is truly human-centered, built for humans based on real human needs.
The story of Siri is full of great achievements of the human mind. It shows us how the power of thought can fuel great technological breakthroughs. It ends with the same man that started it all: von Kempelen, the man with the kind of thinking that gave birth to the first speaking machine, a truly amazing technological achievement. But more importantly, the kind of thinking that creates genuine human experiences.
The Turk’s biggest achievement was to challenge how we think about machines. This is the type of thinking that I like to call design thinking.
Yes, Siri still has its shortcomings, starting with the fact that it’s voice-controlled. But the mechanisms behind it are nothing short of amazing. Properly pairing machine intelligence with true contextual awareness is what created the first conversational interface that actually works.
And simply because it works, it marks an important milestone: It becomes a template for all future voice-controlled interactions. Even Google has updated its interfaces to include conversational and contextual interfaces. What Siri did was show the world a bright idea and made it stick.
More importantly, for professionals, the story behind Siri offers valuable lessons in true experience design, vital lessons in times clearly dominated by form instead of content, where an excessive preoccupation with formalism can impede further developments.
Experience design is more than numbers, boxes, and diagrams. It’s emotional, invisible at the time of inception, innovative, developed intelligently, and deeply contextual. A complex multiplex, feeding on a variety of different disciplines, such as neuroscience, psychology, linguistics, logic, biology, social sciences, computer science, software engineering, mathematics, and philosophy.
Much in the same way that Siri forges new tools from old technologies, good design feeds on AI for the raw materials to conquer human experience. To add function to experience. To add personality.
Avoid fields. Jump fences.
Disciplinary boundaries and regulatory regimes are attempts to control the wilding of creative life. They are often understandable efforts to order what are manifold, complex, evolutionary processes. Our job is to jump the fences and cross the fields.