I Talk to Robots While Driving by @ttunguz
Over the weekend, I found myself in an hour-long conversation during my drive with an AI.
We jumped from discussing Cooper Flagg’s basketball stats at Duke to comparing Carlo Rovelli and Brian Greene’s competing theories of physics to talking about the history of San Francisco. In a daring feat of economic analysis, I asked it to calculate if the after-tax returns of two ETFs were statistically significant & to compare the energy portfolios using the 13-Fs of a few hedge funds for investment ideas.
When a friend hopped in, nothing changed : a three-way conversation where one of the participants happened to be a savantic robot.
When I have an idea now, I don’t scramble for a notebook – I hold the right-option keyboard key to activate Whisper, an AI that listens and transcribes everything I say, even my messiest, most ill-formed thoughts.
The text of these rough ideas are fed into another AI that organizes them, structures them, and helps me shape them into something coherent through conversation.
To create this post, I used a cascade of AIs.
I needed the AI to write like me. I’ve tried fine tuning, but that didn’t work, so instead I fed some of my previous blog posts into an AI to generate a prompt. “Make a prompt to write like this”, I asked. The prompt is about 500 words.
Then I spoke to the computer who transcribed my thoughts, I pasted them below the prompt, and asked the computer to produce this post.
I went back and forth with the machine: “Remove the headers,” “Make it sound more friendly,” “Talk about the stories in the first person.” “Vary the sentence length and the paragraph length. Ensure two paragraphs have only a single sentence.”
The AI’s memory has become a living document. I can work on a piece for a few days, returning to with a new nugget, “I found this new quote” or “I found this new data point. Weave it in.”
Yesterday, I decided my blog theme needed a refresh. Instead of diving into code and wrestling with JavaScript, I screenshotted my current blog and showed it to an AI.
“Reformat this to resemble The Wall Street Journal,” I said. Then The Economist. Then The New Yorker. Pretty quickly, it generated a demo website using my uploaded blog that allowed me to click between the four different themes.
Recently, I read about an engineer who automatically screenshots his inbox and sends those images to a large language model that drafts his emails. It’s an approach to solving a computer science problem I never would have considered. Why integrate with an API if the computer can see?
The hard work of the next few years will be reimagining these workflows. Our tools have fundamentally changed, and so too must the way we work with them.