The Curious Companion: Ep. 19 – Do You Actually Know How to Use ChatGPT?​

Curious Reader!

Welcome to this week’s Curious Companion newsletter.

What you came for is below, and you can CLICK HERE to listen to the episode if you decide you prefer earbuds to eyeballs.

Happy reading!

In this episode I break down what people often misunderstand about ChatGPT and what it can actually do. I walk through the hard technical limits, the most common errors users make, and then share simple ways to become more fluent so you can get the most out of the model. This episode also covers recent updates from OpenAI, along with why sometimes the best use case for ChatGPT is…not using it at all.

The Inspo

This episode is not at all intended to be a stab or a dig at you. My guess is that you do know what you’re doing, BUT I’ve heard some things from some people that also has me like…wait a minute, WAIT A MINUTE, ChatGPT absolutely CANNOT do that…but it’ll still give you an answer, and the answer will be shit, and you might think that ChatGPT is shit. So I decided to make an episode about it.

The other impetus for this episode was an article I read (I be reading lots of things – don’t forget about Curious Finds on my website) that spoke about this idea that creativity follows fluency, and I do very much agree. You can’t solo on the sax until you know how to play it. (I played sax from age 8-18, so that’s today’s analogy.)

So I want to use this episode to help out with that ChatGPT fluency.

A Quick Rundown of Recent Updates

OpenAI has significantly slowed down the speed and magnitude of their rollouts, but they have released a few things over the past few weeks, so I want to do a quick rundown of those just to keep you in the know and then we’ll chat about ChatGPT fluency.

The updates:

• GPT-5.1 rollout
  • I do think they ultimately need to unify the models so you’re not picking which variant to use.
  • This version did get good reviews for its writing capabilities.
  • Also comes with new tone and personalization controls which I first discussed in episode 9.
• Group chats feature inside ChatGPT
  • Like the name implies, you can chat with friends inside of ChatGPT.
  • Use cases would be something like planning a trip or collaborating on a project – meh.
  • Only available in certain countries and I don’t think the US is one of them.
• ChatGPT Go
  • ChatGPT Go, OpenAI’s cheapest ChatGPT tier, rolled out to 8 additional countries in Europe, bringing the total coverage to 98 countries.
  • This is totable because I think it’s only a matter of time before they roll it out in the US and really push the free users there by throttling the free tier and making the experience unpleasant. (Gonna circle back on this free tier point in a bit..
• API/library updates to support the new models
  • Don’t worry about this, though expect a future episode about wtf APIs are.

Do You Actually Know How to Use ChatGPT?

Alrighty so, do you actually know how to use ChatGPT or are you asking it to do shit that it cannot do?

Here’s a list of hard technical limits, aka things that ChatGPT CANNOT do, that I think you might actually care about:

• Cannot open or fully render most URLs
• Cannot access anything behind a login
• Cannot watch videos in the human sense (only parses frames you upload)
• Cannot listen to audio without a transcription step
• Cannot access real-time, unpublished, or private data
• Cannot click buttons, scroll, or interact with your OS
• Cannot bypass paywalls or CAPTCHAs
• Cannot remember past chats unless memory is enabled and intentionally stored

Remember, ChatGPT will absolutely make shit up; very rarely, almost never will it say “I can’t do that.” It will just hallucinate an answer.

The 6 Biggest Errors I See People Making

1. Asking ChatGPT to go to a website or URL and look at something

ChatGPT literally cannot do that and it will 100% make something up. The only way you can get around this would be to use Atlas, OpenAI’s browser, and then use that ChatGPT feature in the upper right corner. A much easier solution is to just copy the text from the page and paste that into ChatGPT.

2. Using ChatGPT for pictures/graphics

ChatGPT just really is not designed for this. You can use it to get an idea about something but if you’re really looking for quality then I’d use ChatGPT to find out what AI resources are out there that actually specialize in that. Midjourney is definitely going to be one you want to check out.

3. Poor prompts — not enough information or context

I’ve been saying since episode one (and my entire movement career) the better the input the better the output. General questions will get general answers.

Example: Is this paragraph good?
​What does “good” even mean?

Solution: Ask ChatGPT to help you achieve your goal.

Example: I’m introducing a new product to my audience that I’d like for them to check out and maybe buy. Do you think this paragraph will accomplish that? What could I change and why?

The other thing to note here is reverse prompting. Show ChatGPT the ideal final version and then work backwards if you’re looking to create a prompt or instructions you’re going to use over and over. Alternatively, if you’re not trying to create a reusable prompt, just use that ideal final version as part of your prompt so that ChatGPT knows what you’re wanting.

4. Fighting with ChatGPT

We’ve all done it. I just want you do to it less. ChatGPT is a probabilistic model, so each iteration you ask for will not be the exact same as the previous, and sometimes the changes make you want to pull your hair out. Your best bet, in general, when things start going off the rails, is to start over and start a new chat.

New chats are also your best bet once the thread gets super long. It has been my experience that ChatGPT starts to kind of glitch, or freeze when I’m working with significant amounts of text within the same chat, and starting a new chat always fixes this.

ChatGPT has memory, and especially if you’re on the paid version, you’ll be fine starting a new chat. (Don’t forget about the value of Projects.)

5. Thinking it’s smart

It is not alive.
It’s not aware.
It’s not sentient.
It’s not thinking.
It’s not reasoning.
…it’s math.
​Really complex math, but still, math.

Action item here: Check its work!

Don’t take everything at face value, and don’t think that this computer program is god. When in doubt, ask ChatGPT to show its work.

A helpful prompt for this:​
​ “Can you show me the exact source of this information? If you cannot verify it, say ‘I don’t know’ and explain which parts are uncertain or inferred.”

6. Using the free tier and expecting ChatGPT to actually remember things across your sessions

On the OpenAI pricing page it says the ChatGPT free tier has “limited memory”, and honestly I haven’t tried it personally (been paid pretty much since day 1), but based on what I’ve heard from others…it is le trash. Whiiiiich makes me think the other features on the free tier might be equally shitty and subsequently the responses it generates might just objectively be worse than on the paid tier.

No, I’m not telling you to give OpenAI your hard earned dollar. I don’t work for them, they ain’t paying me for this.

But:

1. I do think OpenAI will throttle things (aka gradually decrease the quality of the outputs and overall experience) and try to get people onto a paid subscription at some point. They’re losing too much money not to.
   ​
   ​
2. I really do view AI as a bit of an equalizer, particularly in the information space and in the online biz space, and in the places I “hang out”. With the help of AI, one person can learn and do so many things without spending eleventy billion dollars or hiring some huge team to start off with, so I want folks to really be able to take advantage of this technology. That is why I care about the quality of the outputs of the free tier vs the paid.

How I Used ChatGPT This Week

Each episode I include a section where I briefly discuss how I used ChatGPT that week.

Today I want to chat about how I used ChatGPT when I was in Hawaii 2 weeks ago…and that was, pretty much not at all!

Like I’ve said before, my goal isn’t to get everyone using it; I honestly don’t care if people use it or not. When I’m working, when I’m trying to fix stuff around the house, when I need to MacGyver a meal at home, there are use cases. But I don’t view ChatGPT as some infinite-use, universally applicable, thing.

Lex and I were in Oahu for my best friend’s wedding, and the only times that ChatGPT really came up was when I was planning the trip, and then twice when were there.

• On the Jurassic Park tour there was a fruit and we didn’t know what it was and one of the guys on the tour with us was like, “I’m gonna ask ChatGPT!” But…he couldn’t because there was no wifi (for what it’s worth, Apple has that identify-a-plant feature built in, no ChatGPT needed).
• And then I used it when I was trying to plan our walk over to the actual Turtle Beach and wondering the best time to see turtles (spoiler, we didn’t see any).

I wasn’t working in Hawaii so I didn’t use it nearly as much as I do when I’m home.

The tech is super cool, but, like I say on my website:

“I also believe that AI has tremendous potential to help us live more fully human lives.
Whether that means more time with your family, more ways to express yourself, or simply more moments to do what you find fun and meaningful… that’s up to you.”

I was already with Lex, celebrating my best friend’s wedding, experiencing tons of fun and meaningful moments…so, no AI needed.

Da Wrap-up

Alrighty, that’s it for today’s episode. Hopefully you learned a few things that will improve your ChatGPT fluency, and decrease the likelihood of you getting shit answers and thinking that ChatGPT is a POS.

​

As always, endlessly grateful for you and your curiosity.

Catch you next Thursday.

Maestro out.

AI Disclaimer: In the spirit of transparency (if only we could get that from these tech companies), this email was generated with a very solid alley-oop from ChatGPT. I write super detailed outlines for every podcast episode (proof here), and then use ChatGPT to turn those into succinct, readable recaps that I lightly edit to produce these Curious Companions. Could I “write” it all by hand? Sure. Do I want to? Absolutely not. So instead, I let the robot do the work, so I can focus on the stuff that I actually enjoy doing and you get the content delivered to your digital doorstep, no AirPods required. High fives all around.

Did someone forward you this email?
​Click here to join the ChatGPT Curious newsletter family.​

Stay curious.​​