GenAI tools like ChatGPT can act as reflective coaches, which can help us understand any semi-formed ideas that are bubbling away in our brains and test the limitations of our assumptions
I find the idea of using GenAI as a reflective tool interesting, but have reservations. I am a software engineer and I simply don’t trust it. Could GenAI help me challenge that prejudice? Possibly. Perhaps I should ask ChatGPT… or maybe not …🤔
Hey Jim. Give it a try and tell us how it goes! In my experience, ChatGPT is much better at "what do you think of this?" than it is at "tell me stuff."
Hey Rupert. My post is live. You can find it here: https://substack.com/home/post/p-158760238 This took way more time than I had planned, but hopefully I've captured some useful information. Enjoy!
Very interesting. 🤔 Also, do they dream? 🤔🤔 With advances in non-digital (quantum) computing on the horizon these scenarios could be possible … What could go wrong? ….
Another essay to temporarily take my mind off the systematic demolition of the Republic might be just what I need.
Thank you for the kind words and the restack. Now maybe more than my other 2 followers will see it.
As my schedule allows, I am having a series of conversations with ChatGPT about ChatGPT and what is involved in processing my prompts and formulating a response. My goal is to gain an understanding of the limitations of the technology and possible sources of biases in the responses generated. I am also researching outside technical sources directly so I can be as concise as possible in my ChatGPT prompts.
When I am finished I intend to write a summary these conversations, what I have learned and how it alters and/or nullifies some of the pre-conceived reservations about using ChatGPT as a sounding board.
In the meantime here is a link to a particularly interesting conversation ChatGPT's ability to evaluate disinformation and bias in the external online sources it references when answering a prompt on a topic:
Hey Jim. I tried a role play where I played a prickly Flat Earther. ChatGPT's ability to humour me while gently guiding me back to a more reality-based view was excellent! You can have some fun with things like this. The trick is to explain the roles before you begin and then have a "safe word" to go back to normal.
Hey Rupert. I was just working on the section of my essay (post, whatever) regarding ChatGPT's Preferences memory, and I thought about your Flat Earther session. Are you aware of the Preferences feature and how it works? If not, you may want to check what ChatGPT has saved about you after your session. It might just have noticed that you are a "reformed" Flat Earther. You can ask Chat to tell you what it has saved, or you can see it in Settings > Personalization > Manage Memories. In the future, you can use the Temporary Chat option, which is sort of like the Safe Browsing option in a web browser.
I have my own conspiracy theory role-play planned around the topic of the New Jersey drone sightings. I'm a licensed drone pilot, so I know a few things about the misinformation that was in the news about this. This session will be in my posting as an example.
I have spent the entire week having conversations with ChatGPT and (spoiler alert) I am not so much of a skeptic anymore about using it as you suggest in Sounding Board. My planned "summary" of my findings is turning into a full blown Substack post (my first). My goal is to offer some guidance and recommendations, based off of my experiences, on how best to using it as a "sounding board".
I hope to have completed this weekend. I will drop a comment here with a link if you are interested.
While I prepare for my conversation with ChatGPT, I thought I would share this 17 minute video which very clearly and simply explains how it all works. I will be using this as a reference when discussing my reservations about GenAI with ChatGPT.
Thanks for your reflections! The style of ChatGPT is painful to read, honestly. By the way, the link to "comments" from the email takes me to this old article: https://sharpenyouraxe.substack.com/p/having-an-axe-to-grind-sharpening?utm_source=substack&utm_medium=email, for some reason.
Thanks! I deliberately link to an article about why the comments trouble me just in case any CAT nats are thinking of spamming me with nonsense...
Well played!
I find the idea of using GenAI as a reflective tool interesting, but have reservations. I am a software engineer and I simply don’t trust it. Could GenAI help me challenge that prejudice? Possibly. Perhaps I should ask ChatGPT… or maybe not …🤔
Hey Jim. Give it a try and tell us how it goes! In my experience, ChatGPT is much better at "what do you think of this?" than it is at "tell me stuff."
I will do that and post the “conversation” here.
Hey Rupert. My post is live. You can find it here: https://substack.com/home/post/p-158760238 This took way more time than I had planned, but hopefully I've captured some useful information. Enjoy!
Great essay! And thanks for the hat tip!
🙏
If you want another experiment, I think this is going to be (or should be?) a major use case of Gen AI: https://sharpenyouraxe.substack.com/p/when-machines-hallucinate (I also discuss the risks of spam here: https://sharpenyouraxe.substack.com/p/gen-ai-to-flood-the-zone)
Very interesting. 🤔 Also, do they dream? 🤔🤔 With advances in non-digital (quantum) computing on the horizon these scenarios could be possible … What could go wrong? ….
Another essay to temporarily take my mind off the systematic demolition of the Republic might be just what I need.
Thank you for the kind words and the restack. Now maybe more than my other 2 followers will see it.
Wanted to post a progress update...
As my schedule allows, I am having a series of conversations with ChatGPT about ChatGPT and what is involved in processing my prompts and formulating a response. My goal is to gain an understanding of the limitations of the technology and possible sources of biases in the responses generated. I am also researching outside technical sources directly so I can be as concise as possible in my ChatGPT prompts.
When I am finished I intend to write a summary these conversations, what I have learned and how it alters and/or nullifies some of the pre-conceived reservations about using ChatGPT as a sounding board.
In the meantime here is a link to a particularly interesting conversation ChatGPT's ability to evaluate disinformation and bias in the external online sources it references when answering a prompt on a topic:
https://chatgpt.com/share/67c70a82-18c8-800e-8be8-72c27ab6af68
Hey Jim. I tried a role play where I played a prickly Flat Earther. ChatGPT's ability to humour me while gently guiding me back to a more reality-based view was excellent! You can have some fun with things like this. The trick is to explain the roles before you begin and then have a "safe word" to go back to normal.
Hey Rupert. I was just working on the section of my essay (post, whatever) regarding ChatGPT's Preferences memory, and I thought about your Flat Earther session. Are you aware of the Preferences feature and how it works? If not, you may want to check what ChatGPT has saved about you after your session. It might just have noticed that you are a "reformed" Flat Earther. You can ask Chat to tell you what it has saved, or you can see it in Settings > Personalization > Manage Memories. In the future, you can use the Temporary Chat option, which is sort of like the Safe Browsing option in a web browser.
I have my own conspiracy theory role-play planned around the topic of the New Jersey drone sightings. I'm a licensed drone pilot, so I know a few things about the misinformation that was in the news about this. This session will be in my posting as an example.
I'll check it out! I do find that ChatGPT seems to remember absolutely nothing whatsoever. Even the tone goes back to neutral and boring!
It does sound like fun!
I have spent the entire week having conversations with ChatGPT and (spoiler alert) I am not so much of a skeptic anymore about using it as you suggest in Sounding Board. My planned "summary" of my findings is turning into a full blown Substack post (my first). My goal is to offer some guidance and recommendations, based off of my experiences, on how best to using it as a "sounding board".
I hope to have completed this weekend. I will drop a comment here with a link if you are interested.
I’d love to see it!
While I prepare for my conversation with ChatGPT, I thought I would share this 17 minute video which very clearly and simply explains how it all works. I will be using this as a reference when discussing my reservations about GenAI with ChatGPT.
https://youtu.be/2IK3DFHRFfw?si=RVcMOejBaLg63K_K
Did you read Superintelligence by Nick Bostrom? I have a feeling you might enjoy it!
I haven’t, but will put it on my list— next up after I finish Age of Acrimony. Thanks!