Teaching New Dogs Old Tricks

Associative trails – “lists or chains of documents joined together” (Nelson 86)

When creating associative trails within computers (we used the example of file structures in class), the question arises of whether said trails should be intuitive to the user or to the computer. Is it possible to satisfy the minds of both?

As children, one of the first skills we learn is how to group and order things. Sort the books in alphabetical order. Place the blocks into groups of colors. Learn the days of the week, and months of the year. It seems only natural that we might try to organize our computer files in ways that make sense to us, based on how we’ve sorted things in the past.

It gets trickier, however, when there aren’t just a dozen names to sort alphabetically, but ten thousand. Fortunately, computers are great at performing rote operations like these. But what about when we want to sort more complex objects in a more complex way?

Right now, my phone’s memory card is filled with hundreds of photos I’ve taken over the course of several years. My phone can sort them by date, and some – the ones with geotags – can be grouped by the location that they were taken. And that’s about it. Contextual features like “friends”, “selfies”, “school”, and “food”, while instantly recognizable by my own eye, are a great deal harder for a computer to pick out.

Which isn’t to say that there haven’t already been efforts made to teach computers to think more abstractly. Google Photos, for example, does its very best to automatically sort your photos by subject matter. But this is far from  100% accurate, and has already gotten them into a pickle, when the app put a photo of two black individuals into an album called “gorillas.”

It’s obstacles like these that make me skeptical of the roles and responsibilities we can reasonably expect computers to take on. Teaching computers certain tricks, like setting an alarm or shuffling a playlist, is easy. We’ve a while to go yet before we can teach them the things we still struggle to teach our own children. Until we figure that out, computer-generated poetry or AI-professors will remain, for me at least, nothing more than a novelty.

Leave a Reply

Your email address will not be published. Required fields are marked *