A programmer generates code 100 times slower than they type. Does this change for vibe coding?

"I don’t think humans are formal systems. That’s just one working assumption some philosophers and AI researchers adopt." - ChatGPT 5, Aug 2025.

From my morning philosophy chat with ChatGPT - yes, I really do talk to ChatGPT!

Do you think you are a formal system (aka a computer)?

An app to monitor the likely safety of Bitcoin against quantum attack. (c) Valis Labs Inc. Mockup of Dashboard

Bitcoin Threat Level Monitor from Valis Labs
A proposed app to monitor the likely quantum safety of Bitcoin. (This is a mockup. It would take a handful of days to complete.) We monitor addresses that should not change very much.

Is Bitcoin ready for Quantum?
Who would be interested in meeting to discuss next steps in a post quantum Bitcoin world at Bitcoin2025.

Valis Pitch My pitch from Tech Weekend.

Valis Pitch from Tech Weekend, SF.
youtu.be/hiyWxLvt5Nw

Searching for Alzheimer's cures in the Amazon Rainforest. Except from Dr. Alysson Muotri Speech SXSW 2025. He is director of the University of California San Diego Sanford Stem Cell Institute Integrated Space Stem Cell Orbital Research Center and scientific advisor to Valis.

Reconstructing the Human Brain in the Lab, Dr. Alysson Muotri, SXSW 2025 Talk, Austin, TX Dr. Alysson Muotri talked about the human brain at SXSW, including the development of treatments for Alzheimer's, Autism, Rett syndrome, and space aging.Vali...

Reconstructing the Human Brain in the Lab. Dr Alyssson Muotri - UCSD & Valis Scientific Advisor talks about Brain Organoids at SXSW 2025.
youtu.be/BlIOWuQWkw8

State of the art in Biocomputing, piloting a robot with human brain organoids. Except from Dr. Alysson Muotri, Speech SXSW 2025. He is director of the University of California San Diego Sanford Stem Cell Institute Integrated Space Stem Cell Orbital Research Center and a scientific advisor to Valis.

Deep Tech Week New York Physics of Consciousness

Physics of Consciousness: Deep Tech Week New York Panel with James Tagg, Stuart Hameroff, Joscha Bach, and Suzanne Gildert.

Announcement: My Vibe coding paid off! We got our first result for Objective reduction of the wave function. Press release and paper at my website.
valiscorp.com/

Some questions for an Artificial Super Intelligence or an alien visitor.
Do we have free will?
Is the brain a computer or a quantum device?
Will we ever abolish war?
Is there life after death?

Some questions for an Artificial Super Intelligence or an alien visitor.
Can we cure all disease?
Can we clone dinosaurs?
How should we run planet earth?
Will we soon live forever?

These lists come from my book Are the Androids Dreaming Yet.

What questions would I ask an Artificial Super Intelligence (ASI), or perhaps an alien if one ever turned up on CNN for an interview?

Is there a theory of everything (TOE)?
Are we living in a simulation?
Will we live on Mars?

x.com/skdh/status...

Qualcomm Institute UCSD, Photo by Stuart Volkow

Looks like I'm the face of the Qualcomm Institute Innovation Space! This was us building the World Mobile Antenna for deployment on an aerostat.

1. Run simulation only 2. Run on IBM quantum hardware 3. Run both simulation and IBM hardware 4. Display sample circuit 5. Retrieve results from a previous IBM job 6. Get noise parameters from IBM backend 7. Debug 300μs delay point 8. Exit Enjoy...

One of My Bigger Program Menus showing you the breadth of things you can do. === Quantum Dephasing Experiment === Default parameters: T1=300.0μs, T2=150.0μs, Gate Error=0.01 Please select an option:

And I can do it all myself. I don't need to find a Python programmer. James Tagg.

It gets the heart of Quantum Computers looking at the gate gate error, etc. Absolutely amazing, and 9 months of work plus hiring a postdoc if I had done it even 4 weeks ago. All this does take hours of work but no longer months.

Eventually, you will become interested in the details of Qiskit circuits. That's the business end of things. But this way, you learn ONLY Qiskit, not all the tedious support code. My big programs let me do many things, and the above process wrote it all... I wrote not a single line.

Cursor will then tell you what to delete, but it can't delete big blocks itself, so then you have to go in and do a bit of pruning. You don't need to know what you are doing, but you can't really help learning as you go.

Watching it write lots of code, you do learn the language, you can't really help it. If it gets bloated and loses its mind (this sometimes happens), you tell it exactly: "You bloated the code and lost your mind; please fix it." I'm English, so I can't help saying please.

Sometimes, I have to read the code, but it's rare. If a variable needs to be set, like the T1 time or the array of test times, it's sometimes easier to manually edit it. That way, I can pretend I know how to program Qiskit in Python!

I did not yet work out how to use cursor rules, I just put it in the program's comments. But I know Cursor Rules is the better solution I am looking for.

That means they get the setup of Qiskit wrong and keep making the same error. I now have a bunch of rules to keep it all on track. Even Granite-Qiskit from IBM has this problem. I will publish my rules file sometime to help other people.

I recently got access to the IBM Granite-Qiskit version and am playing with that. The biggest problem with ALL the models is that Qiskit updated its APIs twice in the last two years, and they train on Github et. al., and most of the code is pre-update.

In agent mode, I think it uses Claude Code, and it recently got an upgrade with some really good Qiskit data. Often, I use ChatGPT 4.5 as my physicist, and Grok is like the professor I will send the whole code base and error to, ask what the problem is, and then cut and paste its recommendation back

Quantum Vibe Coding
Vibe Coding Means I wrote no code, I did not even read it. I just give it the objective. If it gets it wrong, I tell it the problem and ask it to try again. I use Cursor as the front end and usually Claude 3.7 as the default AI.

Here at SXSW talking about Communications, Quantum Mechanics and Bio Computers. Here until Thursday. Who else is around?

Episode 107: Quantum Startups, Conscious Machines, and Collapsing Wave Functions with James Tagg Entangled Things · Episode

My interview with Entangled Things talking about Brains and Quantum Mechanics.
open.spotify.com/episode/6Ho...

Simulation versus real hardware for a test of Orch-OR Collapse. Shows some interesting differences which need more controls and IBM credits to develop further. Ignore the curve fit lines they don’t make sense yet.

Vibe Coding Quantum Gravity: I asked ChatGPT 4.5 and Cursor to implement my garage quantum gravity Orch-OR experiment on the IBM cloud quantum computer and it said yes! and wrote the code. I ran out of IBM credits but there is certainly a departure from the simulation at short time scales.