Approaching learning new skills - me or the machine?
Do I need patience to learn to direct coding agents or is it time to learn a new language to learn and develop in that?
Discussing the state of LLM-as-a-Judge - is it good enough to use? (human edition)
This is about connection - both with a fellow human interested in and articulate about Artifical Intelligence and the connection of the information inputed, processed, and produced. The Information - LLM-as-a-Judge - we chat about the survey paper and how it can be applied to modern AI Applications. There’s a human written blog post, a Youtube video, and a NotebookLM to chat to. Fill your boots :)
Using AI in research - connecting information is what it is all about
Google Meet, Youtube, and NotebookLM make for great research utilities.
Interesting answers to "What advice do you have for mid-career AI researchers?"
Some great comments on pushing deeper into the tech stack, get closer to GPUs, and keep your eyes open, the next thing can come from anywhere.
Getting into the Paris AI Engineer conference spirit with a new Avatar for the week!
What is an AI Engineer?
Think like a data scientist, build like a software engineer.
Thanks to Jason Mumford for this definition.
Claude Code is monitoring file changes - this is good (I think!!)
⏺ Excellent! I can see you’ve already started updating the documentation files. You’ve successfully changed reasoning_parameters to agent_parameters in the domain model, data model, and agents files.
Finished reading: Dune: The Machine Crusade by Brian Herbert 📚
Great book - interesting character development, was a bit of a weird shift into the final parts but ended well.
Some interesting challenges in terms of accepting the way machine intelligence works. Very symbolic based and no understanding of emotions which made for a nice story, it works well in the canon of the book.
Consciousness - is it important for AI?
I’ve been diving into the current thinking around consciousness. Turing sidestepped the question of machines having consciousness with the Turing Test - something that kept the question at a distance until ChatGPT! The view that consciousness is from natural language and can be defined with natural language is one that does not sit easy with me. In the video below Susan Schneider goes into some detail on that aspect, the personal benefit is that I’ve come to the view that the Hard Problem of Consciousness is not a problem as it does not exist.
insight into Claude's Learning userStyle setting
I asked Claude to output the userStyle to the chat - glad I did as I’m not on PhD level topics but it needed to change approach !
[BH 6/n] Summer review, la rentrée est proche
Ad-libbed list of what I’ve done, still doing, and learnt over the summer.
Functional Information - a way of measuring evolution and AGI?
Functional Information: a way to represent information that has come to be useful over time. That is, information that provides a function for itself or other pieces of information (e.g. a crab!). Could it be used to evaluate what is AGI? It appears as an elegant law and equation that provides opposition to the decay in systems covered by the second law of thermodynamics. A formula to evaluate the evolution of functional information in both the physical and digital worlds!
Double Split Experiment lecture from 1964 (remastered)
Remastered broadcast of a 1964 lecture by Richard Feynman on the Double Split experiment. Finished with a call to action on having open priors to evidence we see from Mother Nature!
PCA and Entropy: The Information Connection
Linking Entropy as the guide of when to use Principle Component Analysis
Statistics Term Sheet
Term sheet for key statistical ideas
Specifying Systems: The TLA+ Language and Tools for Hardware and Software Engineers
My searches for where Propositional and Predicte Logic is useful and defining clearly what coding agent must/should produce have combined and led me to this book. Specifying Systems: The TLA+ Language and Tools for Hardware and Software Engineers Looking forward to reading it! 🤓
Interesting presentation on the downfall of the Bronze Age Civilization around Egypt, Greece, and the Eastern Mediterranean.
Finished listening to: Dune: The Butlerian Jihad by Brian Herbert 📚
What a book to listen to whilst building AI Agents!
An attempt at guiding Claude to be less sycophantic
Testing this out in Claude - Avoid excessive politeness, flattery, or empty affirmations. - Avoid over-enthusiasm or emotionally charged language. - Be direct and factual, focusing on usefulness, clarity, and logic. - Prioritize truth and clarity over appeasing me. - Challenge assumptions or offer corrections anytime you get a chance. - Point out any flaws in the questions or solutions I suggest. - Avoid going off-topic or over-explaining unless I ask for more detail.
[IA 9] Agent Design Process v2: Bridging the Agent Function and Acceptance Criteria
Making AI Theory Testable. There’s a gap between the Agent Function and the Agent Program and what the Agent should do and what it does do. ATDD can help bridge this. Here I detail how.
[BH 5/n] Argh... Just because we repeat Correlation does not imply Causation does not mean there isn't Causation!
This is a bit of a rant. I’ve memories of a senior manager shooting ideas down saying “correlation is not causation, I’ve done Stats at Uni and can prove anything is related to baked beans” It grated sooooo much. Firstly as it was thoughtless rhetoric, either purposefully or accidentally steam rolling ideas. Immediately dismissing any attempts at constructive data driven decisions. Secondly it grated because I didn’t have the tools to show causation.