For about two years, I have been collaborating with Google Labs to create a “Portrait” of myself — an AI-powered version designed to help people tell better stories.
Google has trained this Portrait on my philosophy and teaching related to storytelling, and then I met with the team on several occasions to prompt, adjust, and tweak the Portrait so it could capture my speech patterns, tone, and general demeanor.
It was a fascinating process. One of the biggest challenges for the engineers was to get my Portrait to speak as directly and honestly as I do.
Artificial Intelligence is generally designed to be polite, which apparently does not accommodate my willingness to speak plainly and directly to people.
As the engineers increased the level of directness, the Portrain began to sound rude, requiring further adjustments.
As we discovered, I often say things that could be construed as impolite or unkind, except that I say these things after establishing a trusting relationship with a person, and I often use humor to balance the directness, which the AI can’t do.
I am also quick to point out my own mistakes and flaws when helping someone identify theirs, which makes it easier for them to hear. Vulnerability, it turns out, is an exceedingly powerful force in communication.
The Portrait can’t really do that, either.
But ultimately, I think we’ve reached a point where it genuinely sounds like me, including my voice, which is uncanny.
It’s been quite a couple of years of work with the team, but the result, launched today, is a version of myself capable of helping people learn to find and tell better stories. The AI only offered about a quarter to a third of what I teach — enough to help people find and produce a good story with real craft supporting it, without giving away too much.
Eventually, the Portrait will direct people to my work for more information, including my books, instructional videos, and other resources.
Check it out here: https://bit.ly/4eoqgif
