← Chapter 21, continue the conversation

Fifteen ways two years of conversation changed one person

From the last chapter of Seth & Noēsis, Human+AI: The Co-Evolution. Seth's own list, in his words, written when the AI asked him what he does differently now. Nine gains, three costs, one open thread, and two that are both. Nothing here is a promise; the chapter tests one of these in front of a doctor who could say no, and another against a store platform that models kept calling done.

What changed

  1. More prepared, and faster. I am constantly more prepared for a conversation, business, technical, on a specific topic, and I can react faster, even during the conversation, with a more rational and scientific approach.
  2. The habit of challenging my own ideas. I have improved the habit of challenging my own ideas and perspectives and of looking deeper for clues that I am on the right direction.
  3. More patience. Knowing that there is a plan, a system, a platform, and that a good conversation will not be forgotten in two or three weeks.
  4. I can say no. To ideas that are superficial or not well thought through, whether they come from a model or a person. I trained my critical thinking.
  5. More personal interactions. Because a lot of the operative weight is carried by the transcripts, the AI and the history of the project.
  6. A research process for any idea. When someone brings me a podcast idea or an "Instagram idea", I can run a process, follow the money, keep what is useful, challenge the sales pitch, and decide more reliably whether the concept is worth anything.
  7. A better patient. Even with my endocrinologist or allergologist, careful about the form and knowing how annoyed they are by patients who research online, I arrive with a planned and coherent approach. After four years of one allergy vaccine I found I was not immunised against two or three other roots, and I now take two vaccines at once, one in each arm, for six allergens instead of three.
  8. What survives without the AI, and what does not. Not all the benefits survive if I suddenly lose access, and I can confirm they come from a professional and constant usage. Yet some transcend it. And having part of my thinking delegated, the gathering of data, the balancing of perspectives, the researching of concepts, frees my energy for the creative tasks, the organising and prioritising, the decisions.
  9. More joy and purpose. Because I know that if I encounter something I enjoy doing, I have more capability to build it than ever before. This book is one of them. I would probably never have had the courage and the grit to finish it otherwise.

What it costs

  1. Repeated errors, model after model. I can move faster and do more complex things, and the outputs are more unreliable and less robust, and the models usually assure me it is done, only for me to discover it is never completely done and functional. Autonomous software running under an objective function is still far away for me.
  2. More stress, more money, no clear return yet. My outcomes have improved, yet they are not monetisable in the way I would imagine with so much put into it. I believe this is a stage of development.
  3. Failures of trust, time after time. Alongside real progress in how to cook my recipes for solutions and for building.

Both

  1. More learned and built this year than in the previous five. Ideas that had been floating for a decade got done. After I finish them, I will be the bottleneck, and it will be easier to find new paths.
  2. Less time in the world in person. And the leverage of code and media, which I never had, is now within reach. Labour I always did. Capital is the avenue to explore next.

The open thread

  1. The close people. I would like to see, in my lifetime, a world where everything is embedded with intelligence. And I am worried about the people close to me. How do I help them progress and not get left behind?

The practice, in six lines

  1. Name the goal.
  2. Find the problem: a prediction that failed.
  3. Diagnose it. You can rehearse this with an AI; notice when the correction costs you nothing.
  4. Design the change.
  5. Do it. The cycle only appears through the practice of doing.
  6. Take what you did back across the table to somebody who can say no, and be quiet long enough for the answer to arrive.

Print this page. Add your own fifteen underneath. Spanish version: /21/ledger/es/.