If you’re using LLM agents for your data science tasks, you may feel like you’re the bottleneck. Agents are computationally superior and fast, but if you’re doing serious work, they’re nowhere near autonomous. They need supervision.
To keep up with the pace, you may have added a monitor or two, and turned to speech-to-text to speed up your inputs.
I did both, but my experiment with speech-to-text was short-lived.
Writing is a form of thinking. When I gave up writing for the fool’s gold of efficiency, I also gave up thinking. So it didn’t work out for me. I was quick to get back to typing.
After all, “writing is Nature’s way of showing you how sloppy your thinking is.” — Dick Guindon
This article details a NYT author’s experiment with Wispr, a popular speech-to-text app. It resonated with my experience and prompted this post.