Hi everyone,

Welcome to another edition of AI versus Marshy.

It’s been a week or so - and I'm kindof glad I haven’t said a single thing about GPT-5.

Today we’re looking at:

• mental health doom-mongering
• a little nugget for the hackers on this list
• one-shotting landing pages and how to stay up-to-date with the webinar release schedule

Alright - lots to get on with so let’s make like a knoll and step on it.

AI versus mental health

I’m passionate about mental health, and supporting people living with complex mental health.

I do a lot of work outside of this newsletter in that space around advocacy and sit on the board for a critical Australian charity in this space called SANE.

They support people navigating all of the very hard things that come with recovering from big mental health setbacks.

So when that world intersects with this world I’m going to find it hard not to say something.

So I got linked this over the week and saw it sourced across a number of different social media channels.

Okay - so let’s talk about the good parts of this:

• Holding a big technology accountable. This is important. Just releasing it at global scales and shrugging with a “that’s just how it is” doesn’t cut it, and I’m glad there are people calling them out. I will continue to do so too.
  ​
• There are new social dynamics and cultural impacts that are going be felt from these changes in technology paradigm and I appreciate the reflection and highlighting of this and there should be more of it. See: the hidden cost of AI in Africa​

But I hate it.

AI is not inducing psychosis.

The doctor she Tweet screen capped (but didn’t link or attribute to) said this later in his thread.

And then had to do more media to clarify and make abundantly clear that AI doesn’t cause psychosis but can amplify existing vulnerabilities.

Psychosis is a symptom.

It can be a symptom of too much drug use, a manic episode, a postpartum depresssion etc.

Then the sensationalist language is also fraught.

Conflating a psychotic episode (which will affect around 0.5% of Australian adults each year) with lots of problematic cases of behaviour that’s disturbing isn’t fair.

One doesn’t lead to the other or vice versa.

It just further stigmatises the language used around mental health and unfortunately that’s the nature of an online attention-grabbing play.

The author promotes their newsletter (hey - don’t we all) and also pushes you towards their AI governance training.

I’m not at frontline but a psychologist friend is - they also have examples of in-patients describing delusional behaviour as a result of their interactions with an LLM (something about being they’re an ancient gypsy) and this is common.

But this applies to all technology use.

20+ years ago I was hospitalised, and believed the TV was talking to me (which I recall in my first book):

It’s unfortunate that the recursive loop of a “chat” can lead people towards believing things that aren’t real - but that’s less something specific to AI, and more something specific to broader societal issues (loneliness, inequality, gender violence).

For the hackers (all 3 of you)

I know there’s at least three people that will appreciate this and so if you don’t want a way to create free AI API calls in your dev environment then keep scrolling on…

Still here?

Okay, so one of the benefits of being an absolute hound for growth/marketing/AI/automation leverage is that my spidey sense tingles very effectively when I see angles that are a little too good to be true but work.

I don’t use Kilo Code but love the idea of free, so checked if I could set it up (I can).

I ran into a couple of errors with the API, but “vibe coded” an image recognition proof of concept in about 5 minutes for this newsletter.

I’ve been playing with different coding tools (Claude Code vanilla, Claudia, Kiro, VSCode w/ Kilo, and Toothfairy AI) and not sure what I like yet but it’s a very good time to be a nerd interested in these things.

I warned you about scrolling on...

One-shotting landing pages

I’m still figuring out how to promote my webinars versus the newsletter, but think I’m getting closer to a solution.

I ran “How to one-shot landing pages and why this is useful” on Wednesday and pulled a good crowd, so have some goodies:

• ​Here’s the recording (minus Q&A) on YouTube​
• ​super prompt for one-shotting a landing page (best used with a premium LLM)
• ​slides from the session​

I’m going to reserve promotion of upcoming webinars to only people who have shown an interest - so if you’d like to be notified about upcoming topics and stay in the loop on those, you can subscribe yourself by clicking this link:

=> Tag me as “webinar” interested so I stay up-to-date on these​

This should help keep things in their lanes.

I’m also building a Skool called GrowTechGood - it’s barely set up but will start building it out over the next few weeks and have had 4 signups already (haha - it’s literally to a holder page and I need to get my ASS in SHAPE).

–

It’s been a fun week on the work front as I met some more founders this week (one for a project, one for coaching) and got invited to speak at DigiMarCon next week.

It’s short notice but love a stage so will be doing my thang next week on “Beyond ChatGPT: Building AI Systems That Actually Scale Client Growth” and sharing it with attendees.

As always if you’ve got questions, thoughts, or feedback on the newsletter I am all ears and love seeing feedback from our readers.

Have a great week!

-Marshy