For all the readers who have taken to the air to ride an airplane to another place, they will understand existentially the problems of boarding an airplane. Of further interest is that boarding an airplane is an exact metaphor for trying to integrate AI with human nature. Mariner depends heavily on an article from the January edition of Scientific American magazine.
It is a difficult process to efficiently seat a large number of travelers loaded with luggage, children, carrying a drink and using a single person aisle, overhead storage, predetermined seat numbers and two or three seats in a tight-space row making it difficult to reach a window seat. Understandably airline corporations have tried different sequences to overcome this jumble. They have tried loading from front-to-back, back-to-front, by class, by seat number and by no seat number. None have sufficiently resolved the jumble.
In 2005 Jason Steffen, an astrophysicist at the University of Nevada, Las Vegas, became captivated by the problem. He turned to his computer modeling skills, which he usually reserved for studying the movement of exoplanets, to find a better way to board.
After hundreds of iterations, he found that the most efficient boarding method was a version of back to front—with a few key twists. Rather than have passengers fill in each row sequentially, it was best to start boarding from the window seats, skipping every other row along the way. Effectively, this means that people with an even-numbered window seat would board first, followed by those with an odd-numbered window seat, those with an even-numbered middle seat, and so on. According to simulations, this approach was twice as fast as the front-to-back boarding strategy and 30 percent faster than random boarding.
But alas, when installed for an actual boarding, known as the Steffen boarding method, it works slightly better than the back-to-front method but hasn’t truly solved the jumble. Unfortunately, real people don’t behave in mathematically ideal ways. A large percentage of passengers do not follow the instructions given at the terminal. If people were expected to board in a predetermined order, they could easily miss their number being called because they arrived late to the gate or weren’t paying attention. People assigned a random seat number based on their spot in line might be confused or dissatisfied with their assigned seat and often fliers find themselves in the wrong line. Steffen’s method allows groups of people to either board together or sit together but not both—a huge drawback for families traveling with small children and groups such as students traveling with a teacher chaperone.
Hasn’t every flier experienced this? It is a situation where logic and mathematics, while being applied to human behavior, misses the mark because human behavior is not bound to follow the equation or, in some context, misrepresents the objective to board normally for some reason, perhaps a paraplegic passenger or simply to leave.
Existentially, this may be the standard experience not only for fliers but for medical doctors, insurance coverage, salary descriptions, health benefits, education certificates, background checks for renters, credit cards, etc. Will the presumptive, if simplistic logic of AI be able to deal with humans whose life is similar to New York City’s rodents? – only in terms of their existential lifestyle, of course. Will turning one’s life over to Walmart suffer the same jumble, getting tapa in their groceries instead of tapioca?
Ancient Mariner