Options are an area I’m always interested in learning more about, especially how others think about risk and structure around them. I may take you up on that sometime.
Exactly. If your AI just agrees with everything you say, you're using it as a fancy autocomplete, not as a thinking partner. The real value comes when you set it up to challenge your assumptions and poke holes in your logic. Best outputs come from treating it like a sparring partner, not a yes-man.
Oh I was thinking about that lately. All models tend to agree unless I ask them not to and challenge my assumptions. Over time they learn and start challenging more. I even felt Calude was yelling at me for arguing I was right 😂😂😂
Brilliant read, Joel and bang on the money. You’ve nailed a growing problem most people don’t even realise they’re fuelling.
To your question: the one part of a profile to update first when your direction evolves? Headline. Always. It’s the front door to everything else, and when it’s clear and current, everything you say after lands more credibly.
Now, that bit about GPT-4 confirming your biases more than a human would? Spot on. In fact, a 2024 study from Stanford showed that large language models gave biased responses in 66% of subjective decision-making prompts — especially when prompted with leading assumptions.
You’ve got me thinking: do you find this pattern more pronounced in leadership teams who use AI collaboratively — or in solo decision-makers who never externalise the prompt design process?
I was guilty of this when I first started using the models. The model's outputs were so convincing that I felt almost rude pushing back. The best training for the model, and when we are tuning our own agents, is the back and forth, the longer conversations.
Good to connect. I’m fairly new to Substack and still settling into the rhythm here. I write about markets and risk, but more from the angle of process, discipline, and the mental side of decision-making rather than predictions or hype.
I’ve spent enough time around markets to know how easy it is to overthink or get swept up in noise, so the writing is partly a way for me to slow things down and stay honest with myself.
If you ever have a moment to look through it, I’d genuinely welcome any feedback on the process. Good or bad — I can handle it.
Just subscribed! I’m in the market quite a bit through options trading and index fund investing. I’m glad to take a look, and don’t hesitate to reach out if you have random questions
Great piece! Way too many people use AI like it's just Google.. We should treat it like a senior engineer who isn't afraid to tell us that our idea is bad (although sometimes it's way off too haha!). The trade off is that it takes more time to prompt and play around with it, but the output will be way more valuable.
Thanks for subscribing — appreciate that.
Options are an area I’m always interested in learning more about, especially how others think about risk and structure around them. I may take you up on that sometime.
Glad to connect.
So true. It’s time to start making our Ai efforts more strategic and accountable
Exactly. If your AI just agrees with everything you say, you're using it as a fancy autocomplete, not as a thinking partner. The real value comes when you set it up to challenge your assumptions and poke holes in your logic. Best outputs come from treating it like a sparring partner, not a yes-man.
Oh wow! Setting aside some time to go through these
🙏🏻
I like the sparring partner analogy! Makes you sharper
I have 2 interesting post on sparring partner.
Sparring partner in Notion AI:
https://solopreneurcode.substack.com/p/how-i-built-an-ai-sparring-partner-in-notion-ai
Sparring partner in WhatsApp:
https://solopreneurcode.substack.com/p/brutal-sparring-partner-on-whatsapp
I always love this. I always love when you confront AI and say: "Hey, don't agree with me on the first try. Use this prompt to disagree with me, etc."
It's extremely useful to ask for:
1. Three different solutions
2. Recommendations
3. Questions to make me think about it
And that kind of stuff
Amazing article Joel!
Oh man those follow-ups are great
Oh I was thinking about that lately. All models tend to agree unless I ask them not to and challenge my assumptions. Over time they learn and start challenging more. I even felt Calude was yelling at me for arguing I was right 😂😂😂
That's so funny. Actually, Claude can get feisty sometimes haha
Brilliant read, Joel and bang on the money. You’ve nailed a growing problem most people don’t even realise they’re fuelling.
To your question: the one part of a profile to update first when your direction evolves? Headline. Always. It’s the front door to everything else, and when it’s clear and current, everything you say after lands more credibly.
Now, that bit about GPT-4 confirming your biases more than a human would? Spot on. In fact, a 2024 study from Stanford showed that large language models gave biased responses in 66% of subjective decision-making prompts — especially when prompted with leading assumptions.
You’ve got me thinking: do you find this pattern more pronounced in leadership teams who use AI collaboratively — or in solo decision-makers who never externalise the prompt design process?
Thanks for sharing, Melanie! Excellent points! I see this within both actually!
I was guilty of this when I first started using the models. The model's outputs were so convincing that I felt almost rude pushing back. The best training for the model, and when we are tuning our own agents, is the back and forth, the longer conversations.
Hi Joel,
Good to connect. I’m fairly new to Substack and still settling into the rhythm here. I write about markets and risk, but more from the angle of process, discipline, and the mental side of decision-making rather than predictions or hype.
I’ve spent enough time around markets to know how easy it is to overthink or get swept up in noise, so the writing is partly a way for me to slow things down and stay honest with myself.
If you ever have a moment to look through it, I’d genuinely welcome any feedback on the process. Good or bad — I can handle it.
Cheers, Andrew
Just subscribed! I’m in the market quite a bit through options trading and index fund investing. I’m glad to take a look, and don’t hesitate to reach out if you have random questions
Great piece! Way too many people use AI like it's just Google.. We should treat it like a senior engineer who isn't afraid to tell us that our idea is bad (although sometimes it's way off too haha!). The trade off is that it takes more time to prompt and play around with it, but the output will be way more valuable.
Thanks Mia, and you are right it does take more effort!