Thoughts on human vs AI narration

My Instapaper feed looks a lot like my Steam account. I have gathered many items with the good intention of making use of them on a rainy day, but they eventually gather dust on their virtual shelves.

Occasionally I’ll try and blast through a set of articles that I considered at one point to have potential value to me. When I’m strapped for time I’ll try Instapaper’s “speak” feature, which is effectively a shortcut to iOS’ dictation feature. I usually end up cancelling the dictation after it reads a few paragraphs. Perhaps I forget how bad it is or maybe I’m expecting some advances in the technology since I last used it that make it sound more natural and human-like.

With AI being applied to more and more tasks, fields and entire industries I notice how far it falls short of key human traits such as empathy and storytelling, and empathy within storytelling for that matter. We have thousands of years of experience on the robots when it comes to communicating oral wisdom from campfires and passing ideas down through generations.

AI narration is improving, however, services like Narro offer bots with various dialects that have fairly accurate speech patterns and rhythm. But ultimately the narration ends up having a passive effect as AI hasn’t managed to emulate the human capability of captivating an audience to deliver a message. Robots rely on punctuation like a musical score to inform the pace at which they communicate text, though we lack punctuation for instinctive human storytelling tools like drama and the pause to allow certain ideas to simmer in the minds of the listener. This is why uninteresting speakers often fail to communicate their idea — the audience becomes disengaged from the message.

On a relevant note, another key human trait in storytelling is context. With a contextual understanding of the topic, humans know when to employ the storytelling tools mentioned previously. Consider this example from a recent opinion piece in the New York Times about the dangers of relying on Google rather than investing in deep understanding:

Google is good at finding information, but the brain beats it in two essential ways. Champions of Google underestimate how much the meaning of words and sentences changes with context. Consider vocabulary. Every teacher knows that a sixth grader, armed with a thesaurus, will often submit a paper studded with words used in not-quite-correct ways, like the student who looked up “meticulous,” saw it meant “very careful,” and wrote “I was meticulous when I fell off the cliff.”

From You Still Need Your Brain by Daniel T. Willingham

A classic example of schoolchildren discovering a tool like a thesaurus and employing the technology incorrectly. With AI dictation we are cutting out the middleman leaving it open for potential missteps like this not necessarily with words, but the in the power of those words.

P.S If you are looking for a startup idea, please feel free to design a subscription service with great human storytellers willing to read my Instapaper feed for me, thanks!