How to Wash Dishes AND Use ChatGPT Like a Stone-Cold Pro

Hey there, I’m Zach Abramowitz and I’m Legally Disrupted.

You might be washing the dishes incorrectly, and it could be affecting your experience with ChatGPT. Come with me to analogy land.

When I got married nearly 17 years ago, I quickly discovered that my wife Penina and I had a different way of doing the dishes. I would take the dirty dishes, stick them in the dishwasher and simply ask the machine to do the work for me. Penina would regularly point out to me that my technique doesn’t actually clean the dishes (85% accuracy rate at best i.e. dirty dishes 🤢). When I asked how it was our plates were all so shiny, she confessed that she would often take the dishes that I had outsourced to the dishwasher and clean them again by hand.

Now, her preferred method, on the other hand, was to rinse and wash the dishes lightly before sticking them in the dishwasher and then let the machine take it from there. The result? Sparklingly clean dishes every damn time.

At the moment, there are two parallel methods for people using ChatGPT: “dishwasher only” folks vs “dishwasher plus” people — for ease of use let’s call them the Zachs vs the Peninas. The Zachs are trying to use ChatGPT as a pure replacement for manual tasks eg “ChatGPT do my job.” The Peninas are using ChatGPT on top of their own work, and the results are amazing. Ironically, when it comes to AI, I am a Penina, not a Zach.

Peninas love ChatGPT. It’s the Zachs that are getting into trouble. The poor lawyer who used ChatGPT to write his brief (filled with made-up case law) which he submitted to the court without reviewing — total Zach. Guy who complains about the results after prompting ChatGPT to “draft me a contract”? such a Zach move.

Peninas better understand the use case for AI. They see large language models as the machine dishwasher that takes pristine and clean to the next level. At the moment, Peninas are still putting in initial work themselves, but less than they had to with pure hand washing (man, I am really not giving up on this analogy!)

ChatGPT doesn’t replace you, but it can help you scale. And the more information you give ChatGPT, the more tailored the outputs. As opposed to Google searches where less is more (too much information could actually lead you away from the right keyword matches), with ChatGPT MORE is more. My own experience is that ChatGPT doesn’t always get it right the first time, but I can’t remember having to ask for more than 3 revisions. I know to ask for revisions and fixes because, like I said before, I’m totally a ChatGPeninaT kinda guy.

If you’ve had lackluster results, ask yourself whether or not you’ve been trying to use LLMs as a replacement or as steroids. If you’re prompting ChatGPT and not getting it right the first time, consider whether you tried giving it feedback or whether you can provide more information/context about the task it needs in order to work its magic.

And whatever you do, stay disrupted, it’s getting crazy out there! I’m Zach Abramowitz and you’ve been legally disrupted.

PS I will be moderating a conversation about AI adoption by in-house legal with some absolute all-stars this Wednesday at 1pm Eastern (check out the banner below). If the conversation we had in our production meeting is any indication, it’s going to be straight fire. Registration is nearly sold out, so click the link and sign up if you don’t want to miss out.

https://www.axiomlaw.com/061423-webinar