Artificially Intelligent Environments

A movie scene that haunts me is one from Steven Spielberg’s Minority Report. John Anderton, played by Tom Cruise, walks hurriedly through a fantastical public street, continuously accosted by holographic advertisements, all of which know his name.

“The road less traveled… John Anderton.”

Moments before, cameras scanned the eyes of passengers exiting the train he was riding, recording their personal identities and locations. And not only did these “idents” go to local authorities, they were apparently distributed to whatever ad network was powering the experience ahead of them.

I often think, if this were indeed a reality, who would have designed such a street? Was it an architect or urban planner hired to maximize the retail sales of a stretch of storefronts? Did they win a design competition for the job or respond to an RFP? Is the entire thoroughfare owned by a single retail conglomerate, tracking all potential customers? And moreover, how many interaction designers knowingly augmented this experience by designing ads served and positioned in realtime? (Or perhaps they are merely AI-generated.)

After all, this already happens to us on the Internet and within mobile apps. We browse; we click; we’re tracked; we’re profiled; ads are positioned in our view, sometimes with exceeding insistence.

Though I hesitate to start with such a dystopian image what I intend to be a positive discussion, it does show the aspects of an artificially intelligent environment, arguably gone wrong. And in a not so subtle way, it exposes an alliance between organized authority and capitalism, so perhaps it’s an experience gone right from the authorities’ and retailers’ perspectives.

It shows how environments can impact the occupants of space, how it forms and crafts a type of space, one of an exchange of identity and attention. And it demonstrates how AI could be a proactive signaler in such spaces.

Intrinsic properties:

  1. Drive on the right side
  2. …

Emergent properties:

  1. Driving on the right side of streets means left turns are costly, even dangerous.
  2. On a chess board, the “center” is a strategically important area of the board, not because the board is designed that way specifically, but as a secondary effect of the rules of the game.

Agents must navigate these properties, both intrinsic and emergent.

Occupants (and Agents)

Spatial agents occupy and traverse spaces.

Three contexts comprise spatial AI:

  1. Designers of spaces. Here, AI helps designers in conceiving of and configuring spatial ideas. This means professions like architects, urban planners, factory designers, etc.
  2. Environments. That is, where AI can imbue environments with novel, sophisticated, and “intelligent” behaviors, influencing and impacting the experiences of occupants.
  3. Occupants of space. This means AI methods occupants may wield to traverse the space they inhabit. This could be obstacle-occupants or agent-occupants.

Environments and Agents

Navigating an airport you’ve never been to has its challenges, but if you travel reasonably often, you know what to look for. There are terminals, ticket counters, security checkpoints, gates, customs, etc. Gates often segregate domestic and international travel. And there is typically a plethora of signage to guide you… but you have to locate the signage.

And often times, we get more sensible cues from other travelers around us, perhaps those who have a familiar knowledge of where to go. Even if we don’t interact with them directly, like asking directions, we can often sense the flow of the crowd or a forming line.

Environments

Go has proximity, nearness, and continuity. It doesn’t really have “distance” however. We don’t really traverse the space of Go like in Chess, where the properties of the agent give it different abilities to traverse locations.

Even more so, what an “environment” could mean is more than our daily physical and temporal experience. It could mean a political environment. Or a social one. Or a virtual one.

Navigating a political arena certainly has its tactics. To do so successfully, one has to have an intuitive sense of others’ motives and sensibilities.

Imbuing Environments with “Intelligence”

Spatial AI considers the mechanisms to imbue our environments with forms of responsiveness in ways to either improve our lives or to constrain them.

And by “environments,” I mean the context or overall situation surrounding spatial “agents” (like us) that provides that “economy of opportunity.” An environment thus holds and contains the sum total opportunity available, like the opportunities to inhabit physical locations over time.

Environments can offer vastly different spatial properties depending on perspective. In a city street, for example,

To control the flow of traffic, city streets employ the aforementioned traffic lights as well as cameras and sensors to help enforce rules and contribute data to a centralized system coordinating the overall state of the street grid.

Thus, “intelligent” environments have the ability to…

Incorporating AI methods like computer vision and predictive algorithms, designers can deploy interactive and adaptable behaviors in environments, which can adjust to and change occupants’ perceptions and the availability of opportunity. Moreover, they can deploy such means not only for practical reasons, like reducing traffic jams, but also to create delightful experiences, like adjusting lighting or music in a space or making way-finding in a complex interior less burdensome.