Max Milne
banner
maxmilne.bsky.social
Max Milne
@maxmilne.bsky.social
Psychotherapist.
All my friends are nerds, but I'm not.
www.maxmilne.com
Reposted by Max Milne
arxiv.org
September 18, 2025 at 8:16 PM
I've been using RooCode for a while with CC + Gemini 2.5 under neath for projects. It was good. Now I'm running ChatGPT via Codex in VSCode. I have had only success with it. For people new to messing around with Code I will be suggesting this mix. Very user friendly:
September 2, 2025 at 3:28 AM
Reposted by Max Milne
i do actually think it is a meaningful milestone -- with rather seismic implications for the social web -- that the median bot is getting smarter than the median internet commenter
August 11, 2025 at 10:24 PM
Reposted by Max Milne
From Memories to Performances. 📷
July 15, 2025 at 9:57 AM
Reposted by Max Milne
I think that's another point in favor of the analogy, personally: industry trying to replace programmers with LLMs is stupid in the same way that trying to replace carpenters with table saws is stupid
July 6, 2025 at 3:19 PM
Reposted by Max Milne
OpenAI is now required by court order to preserve all ChatGPT logs including "temporary chats" and API requests that would have been deleted

if I understand this correctly, it means data retention policies for apps that use OpenAI API simply cannot be honored
June 4, 2025 at 11:21 PM
Loving @obsidian.md (thanks @stephango.com!) & moving all academic research there. iPad PDF annotation was a hurdle (Zotero reigned). New ruler? Obsidian with PDF++ plugin! Made a video on how: www.youtube.com/watch?v=HM_T...
May 2, 2025 at 6:25 AM
*I picked up new "Sky" blue MacBook Air - feels more of the pastel hue edition of "blue plastic film over computer components" from the 90s. Not in love with it. Could warm up.
March 12, 2025 at 7:45 AM
Taking note from @emollick.bsky.social "VIBEcoding" I've been playing with making an @obsidian.md Web Clipper that integrates the Interpreter function to make a 'one click' template for Article capture - News, Journals, Research etc. I wish I had this when completing my Masters. Feel free to play.
March 12, 2025 at 7:31 AM
@strange.website - Thank you for writing the henry.codes/writing/how-... post - just landed a new MBA M4 (yes sky blue) and had heavy lift help from Claude to help make it happen....yes I remembered to not give my SSH keys to the Ai overlord. I'm wondering if Claude Code could run this freehand?
How I Set Up A New Laptop | Henry From Online
The personal site & portfolio of creative web developer Henry from Online
henry.codes
March 12, 2025 at 2:11 AM
Reposted by Max Milne
This is a crazy paper. Fine-tuning a big GPT-4o on a small amount of insecure code or even "bad numbers" (like 666) makes them misaligned in almost everything else. They are more likely to start offering misinformation, spouting anti-human values, and talk about admiring dictators. Why is unclear.
February 25, 2025 at 9:01 PM
Reposted by Max Milne
The trouble is, once I started imagining they were taking selfies I couldn’t unsee it
February 22, 2025 at 4:46 PM
Reposted by Max Milne
All the cloud platforms use your data to train LLMs. Local-first tools are the way to fight back.

Security, defense, finance, healthcare, government... the reason high-security orgs use Obsidian is for the privacy and ownership of their data.

Now every team can be sovereign.
Obsidian is now free for work.

Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore the organizations that support Obsidian on our site.

obsidian.md/blog/free-fo...
Obsidian is now free for work
Starting today, the Obsidian Commercial license is optional. Anyone can use Obsidian for work, for free. Explore organizations that support Obsidian on our new Enterprise page.
obsidian.md
February 20, 2025 at 3:31 PM
Reposted by Max Milne
This new paper shows people could not tell the difference between the written responses of ChatGPT-4o & expert therapists, and that they preferred ChatGPT's responses.

Effectiveness is not measured. Given that people use LLMs for therapy now, this is an important (and urgent) topic for study.
February 15, 2025 at 6:30 AM
"Place the words aside, how would you say that deep thing with sounds and textural utterance alone?" Revisiting VOICE, the 2013 project Adam Hunt and I presented in Edinburgh, Melbourne and had picked up for National broadcast. The body is the score.
maxmilne.com/index.php/pa...
Voice - Radio National - Long Story Short Edit
A 5 minute track edited for ABC Radio National from the sound art Piece called VOICE. Created by Transmit Receive.
soundcloud.com
February 12, 2025 at 4:38 AM
I'm exploring how Relational Meditations fits into my 1-on-1 therapy stack. 1. Help people learn skills to experience, differentiate and articulate felt sense, thoughts and action impulse. 2. Normalise bi-directional process commenting. 3. Then go deep with a shared language and experiential set.
February 9, 2025 at 8:55 AM
meditatewith.ai is a super interesting approach @vincehorn.space and Craig at Interbe.ing took to utilising AI as Social Meditation practice partner. Maybe a bit redundant now with ChatGPT Advanced mode. Though check out the intention, which I write about on my blog: maxmilne.com/index.php/20...
February 8, 2025 at 10:43 AM
While I could practice mindfulness without AI assistance, I found the guided interaction helped my mind settle more quickly into presence. Curiously I could see myself becoming more dependent on the technology...
maxmilne.com/index.php/20...
February 4, 2025 at 6:52 AM
I've recently been involved with www.interbe.ing program for Multi Player Meditation. I'm taking a Counsellors lens to reviewing the program and this is my first take - maxmilne.com/index.php/20...
Interbe.ing
Interbe.ing - Making meditation multiplayer
www.interbe.ing
February 4, 2025 at 6:24 AM
Reposted by Max Milne
A note for instructors assigning research projects (a very effective assignment & key to academia in many ways). You need to understand how far Google & OpenAI's Deep Research tools compromise this work.

You cannot just let learners use AI and assume they will need to fix flaws. It is very good now
February 3, 2025 at 2:56 AM
Reposted by Max Milne
It’s the new American 80s so I dug out my ancient copy of THE CLOTHES HAVE NO EMPEROR, a book which laid bare the Reagan admin just by listing all the idiotic things that happened. Introduction still holds up!
January 29, 2025 at 12:46 AM
Thinking about Telstra Tower and imagining it as a miracle performance art space. One weekend only. Full access. Revolving restaurant. Some how the whole place is falling apart and the car park is more burnt out then space holder of the 360 view inclined.
December 24, 2024 at 11:10 AM