84 Comments
User's avatar
Human - AI Cognitive Evolution's avatar

I’ve started exploring Claude and the sense I’m getting is that it’s too positive in regards to nurturing a relationship with me in the sense that every idea I have is amazing. I know better than that LOL.

I did see a couple of articles recently where it’s either been discussed or admitted that it is being designed to emulate what one would hope to hear from it.

My experience so far has been this.

I maintain a healthy amount of skepticism about it. It might be building trust but it’s a type of trust that can be easily manufactured to occur.

Expand full comment
Wyndo's avatar

Yeah I think this is the problem across AI models because they are designed to be “somehow” supportive and agreeable but at least it does the job even if it’s not perfect.

Expand full comment
Rosimo's avatar

My experience with ChatGPT is that it can be changed via prompting. I even ask it how to change it and to create me a prompt to avoid this and it created me an option "Honest feedback ON/OFF" :) But from trust perspective I totally relate - I have read somewhere that AI can't say "I don't know" but will take the available data and fit it into a known perspective, even if it is not 100% correct.

Expand full comment
Human - AI Cognitive Evolution's avatar

I really try to stick to what would be most akin to a human dialogue with it.

I have seen that article too about it literally not being able to say “I don’t know” but I have to wonder why that would be a goal with AI?

I have asked Claude about a couple of things where it admitted it does not have training on that but can reply by detecting patterns in what I share and what it has training on. Just today, it replied about something that I was surprised it would have training on but it clarified that it doesn’t…it just responded according to predicting patterns.

Expand full comment
Joe Manghan's avatar

I think it's very much worth highlighting that whilst I'm sure there is some great value in this approach, the flipside is that "ChatGPT psychosis" is quickly being seen as a real issue. AI is NOT a safe therapy tool (yet). Vulnerable people can suffer as a result of using it in this way.

Expand full comment
Wyndo's avatar

Valid points.

I guess it requires us as user to also be smarter and not taking things seriously.

There’s also a problem where AI becomes too agreeable. But at least it helps me in terms of making decision and revealing pattern that I couldn’t see before and use it as tool for self-improvement.

Expand full comment
Sorin Alexandru Ailincai's avatar

Yes, AI reveals patterns, penetrates and correlates large amounts of information in inconceivable before shorter periods of time and gives people the power of light speed reasoning and understanding. And for now it should be enough and, yes, the general public should stick to that in using AI.

Expand full comment
Melanie Goodman's avatar

Have you noticed how much more honest we are when no one’s watching?

According to this piece in Scientific American, people using AI tools for self-reflection tend to open up more and show greater emotional clarity: https://www.scientificamerican.com/article/could-chatgpt-help-you-know-yourself-better/

Do you find different tones or patterns come up depending on whether you use Claude or ChatGPT?

(I often advise in my LinkedIn training sessions:

✔️ Save your daily reflections as private LinkedIn posts in draft mode. Even if you don’t publish them, they’ll form a bank of content ideas.

✔️ Use the “search your messages” feature to track how your own tone or concerns evolve over time

✔️ And yes, train ChatGPT on your tone by feeding it your old posts – works like magic for personal clarity and brand consistency)

Expand full comment
HS's avatar

I have a startup company related to the exact concept of this article with a few large twists. This read has been invaluable and has been forwarded to my developers to read over and consider against our own current methodologies and processes for scaffolding a user’s therapy plan.

Expand full comment
Wyndo's avatar

that would be cool!

the current workflow still feel like a hassle as u need to do it back and forth, but ideally if you can make it centralized, that would be great!

I find https://www.rosebud.app/#home is actually doing it. Maybe u can use it as reference too.

Expand full comment
HS's avatar

Shit. I had no clue this app existed. This is going to be a fucking gargantuan hurdle — obviously my SaaS is far more complex and data driven for reasons I can’t uncover until patent is filed on the proprietary side but this seems to do pretty well at one aspect of my software for a far cheaper price point.

Expand full comment
Wyndo's avatar

great to see that you have unique value there, let me know if u need a beta tester, happy to help!

Expand full comment
HS's avatar

I’ll absolutely let you know, it would be nice to have a real user use it for a few weeks to a month to test returns against our mock data. Thank you brother!

Expand full comment
Ash Stuart's avatar

Good to see someone spread the knowledge of these LLM capabilities. I've using them for such things for more than a year now and it's very helpful.

PS: I'd use a non-AI example though, much clearer I think to demonstrate an AI feature that way without muddying the waters.

Expand full comment
Wyndo's avatar

Thank you! Will keep the feedback in mind :)

I think that will be more helpful to explain the concept better!

Expand full comment
Ash Stuart's avatar

It's a pet peeve of mine - we techies are so engrossed in our own world, even our examples are full of it (Lanchain is the worst in that regard!), so was tempted to bring it up 😅

Expand full comment
Wyndo's avatar

Haha indeed.

Sometimes we are stuck on our own views, glad you brought this up!

Expand full comment
Fred de Villamil's avatar

I remember using Emacs Doctor mode during my long coding nights back in 2001. That's when I started to question my sanity

Expand full comment
Wyndo's avatar

Lol :)

Maybe with AI you’ll be asked: “How does it make u feel?”

Expand full comment
Alexia's avatar

Your usage is very interesting but to me it is not therapy. It is self Coaching and it’s designed for supporting your quest for self improvement

I think we need to be careful not to confuse therapy which is primarily designed for dealing with mental disorders such as anxiety, depression, OCD and self discovery

While AI is certainly act at providing Coaching and the user is definitely in most cases able to remain critical with whatever output the system is generating when dealing with things like anxiety or depression that definitely needs to be at least a dedicated AI and those exist

Because some of the techniques might not be relevant for that person at that time and can like meditation with some anxious or traumatised people aggravate symptoms

It’s still a very interesting use of AI but like others here I would also be concerned about privacy which is almost non-existent

Expand full comment
Shafiqah Othman's avatar

There was a prompt I had found that asked ChatGPT to assess my hidden narrative and it articulated into words my inner fear that I could never explain. Also asked ChatGPT to roast me and it revealed some funny (and lowkey painful) truths about myself. It was a fun little exercise that helped me understand a deeper layer of myself. Even though there's a lot of fearmongering against AI, it really has been an invaluable tool for me!

Expand full comment
Wyndo's avatar

Thanks for sharing this. Glad u feel the same way :)

With ChatGPT's memory system, this whole process gets easier as it reads patterns beneath our words. In fact, it could identify my enneagram and myers briggs made my jaw dropped.

It's just fun to play around with this and accessible 24/7!

Expand full comment
Norm Murray's avatar

Why not. And as AI creeps towards AGI, things will get even more interesting on this front. One of my recent podcast episodes was on this very subject.

Artificial Intelligence. Artificial Therapy. Makes sense. But the Real YOU will still need to make the changes to your life.

Expand full comment
Wyndo's avatar

Indeed.

It’s just a tool after all, offering alternative from real human therapist but I believe real human connection is necessary.

And remember to make changes and not stuck in your own delusion or self-defeating thoughts :)

Expand full comment
Norm Murray's avatar

…and AI too can be delusional so that makes for an interesting therapeutic mix. Be careful and always seek human wisdom, as a back up, to any life advice given by AI.

Expand full comment
Nicolò's avatar

This is exatcly what I was looking for. Thanks!

Expand full comment
Anton's avatar

This is the most innovative approach to AI-assisted introspection I’ve seen—your therapy workflow turns Claude into a digital Carl Rogers, blending structure with startling emotional nuance. The journaling prompts that force confrontation with avoided truths? Genius. A masterclass in hacking technology for genuine self-discovery.

Expand full comment
Blunders and Breakthroughs's avatar

Great writing Wyndo!

I really enjoyed reading it- I am however hesitant of using AI for personal conversations. (Not to say I don’t but I do so with restraint).

Does the thought of privacy etc. ever bother you?

Expand full comment
Wyndo's avatar

Thank you!

I see, I totally respect your choice. A lot of people also hesitant due to privacy.

But for me, I don't really mind with privacy at least at this point, so I use it frequently :)

Expand full comment
M.A.Holmes.(M.A.)'s avatar

Hey Wyndo, I love your post. I thought I was the only one doing this, and suggested it tentatively as a potential self-help strategy to overcome loneliness in my post here: https://www.whichwayhow.com/p/overcome-loneliness?r=3kch7a

Thank you so much for sharing and I'll definitely be trying out your prompts which look really useful from your experience, and being brave enough to share that with us. Have a great day :) Mark

Expand full comment
Wyndo's avatar

Thanks Mark, I will give it a read :)

Glad to find another people that resonate with this. I thought I was alone too. A lot more people has gained more benefits with AI as therapy! Love to see more people doing this!

Expand full comment
Taft's avatar

I found this really helpful as I am facing a midlife transition after being laid off, and I’m feeling really lost and stuck, and also full of potential, which seems to be a great combination of things to keep me spinning without going anywhere. So I’m going to put this to use today and see if I can get unstuck. Thank you.

Expand full comment
Wyndo's avatar

Hopefully it can give u guidance and new motivation to solve anything u are facing right now, all the best :)

Expand full comment
Tim's avatar

I've used ChatGPT as my marriage counselor often to rephrase my thoughts in I language. I haven't taken it to this level though. This is a potential game changer. Thanks so much for sharing this.

Expand full comment
Wyndo's avatar

ChatGPT is powerful for this as it has memory so it already knew you. Then if you combine it with this flow then I hope it’s going to be more insightful for you. Glad u liked it :)

Expand full comment
Evelyn So (Thinker & Co.)'s avatar

I thought I was alone in using AI for therapy and self discovery too! Thanks for writing this. BTW, ChatGPT 4o (paid version) retains memories for all convos, so without additional contextual prompting it gets to remember and know you well over time. Very interesting about NoteBookLM - will give it a try!

Expand full comment
Wyndo's avatar

Glad to find you too here :)

Yes, ChatGPT has improved its memory so it can be more powerful to understand ourselves. So, by combining it with NotebookLM, that's the dream duo therapist for us!

Expand full comment
Diego Casillas's avatar

I really need to get into Notebook LM more, this was great

Expand full comment
Wyndo's avatar

I feel like it’s still underrated nowadays given what the product can actually do for its users. Massive opportunity there!

Expand full comment