March 31, 2023
A week-long ChatGPT productivity experiment reveals both its shortcomings and potential.
At the start of this week, I intended to use ChatGPT—running on the latest version, GPT-4—as my daily productivity assistant. I wanted it to help me plan my week, prioritize daily tasks, and find other activities it could assist with.
It wasn’t a great success.
As we were planning my week together on Monday, ChatGPT promised all kinds of support. For example, it would remind me of meetings and keep me on track with tasks.
Of course, it did none of those things.
ChatGPT doesn’t have access to my tasks, and it can’t send reminders. (Though as I'm writing this, it probably can because since last Friday, ChatGPT supports plugins like Zapier—this is the insane speed at which things are now moving in AI.)
Reviewing my candidate tasks for the day and selecting the ones I want to work on only takes a few minutes. Providing ChatGPT with the criteria and context to choose and prioritize items takes much longer. (Though, very soon, when it connects to or lives inside your task management app, I'm sure it's fast and helpful for precisely this.)
By Wednesday, I had given up on involving AI in my planning; I wasn't gaining much from the effort.
While my original experiment failed before the week ended, I discovered three other valuable ways to use ChatGPT.
I identified a range of tasks throughout the week that GPT-4 could help with:
Next week’s article will feature these use cases and include prompt templates for each one.
Because of ChatGPT’s ~18,000 character limit, I could only feed it one week of my time-tracking data from Lifeline, our Pomodoro app. Still, it found some interesting patterns (emphasis in bold mine):
“After analyzing the provided data, here are some patterns and habits that I've noticed, which may be beneficial or harmful:
Higher quality ratings seem to be associated with longer sleep durations. On the days when your sleep duration was 7 hours or more, your self-rated quality ratings were consistently higher. This suggests that ensuring sufficient sleep might be crucial for maintaining high-quality work.
Your productivity seems to peak around mid-morning (10 AM - 12 PM) and early afternoon (2 PM - 4 PM). During these time slots, you tend to have higher quality ratings, which could indicate that you're more focused and efficient during these periods. Scheduling more demanding tasks during these timeframes might lead to better overall performance.
On the days when you had a break of 2 hours or more between work sessions, your self-rated quality ratings were relatively lower in the subsequent sessions. This could indicate that taking long breaks might disrupt your focus and flow. Experimenting with shorter breaks or using a consistent break schedule (e.g., the Pomodoro technique) could help you maintain focus and productivity throughout the day.
On days when you exercise in the morning, your overall self-rated quality ratings seem to be higher. Engaging in morning physical activity might be helping you stay energized and focused during work hours. Incorporating a regular exercise routine into your mornings could be beneficial for your productivity.”
None of these tips are earth-shattering, but they're based on only one week of information. I can't wait to see what ChatGPT will discover and suggest once we connect to its API and load up weeks' or even months' worth of data—something we'll do soon.
ChatGPT, especially when powered by GPT-4, is an incredibly useful writing and content marketing assistant. But it’s also a frenemy, as it’s taking over tasks traditionally performed by human writers like myself.
I’ve launched a new publication called We Eat Robots to help content marketers and writers work with AI. I’ll share my journey—and soon those of others—on learning to work with AI as a writer. I’ll publish interviews, practical tips, and the prompt templates I’m building, testing, and using.
Footnote: If you are wondering or worried about the speed at which AI is developing and what might happen, I highly recommend the episode Freaked Out? We Really Can Prepare for A.I. In this podcast, New York Times journalist Ezra Klein discusses this topic with Kelsey Piper, a senior writer at Vox who has been covering advanced AI for years.