There are days when you are so totally befuddled because your software is behaving badly, you have to bring out the heavy guns, namely, your gigabit-sample-per-second dual channel oscilloscope. This was one of those days.
Grady Booch
@booch.com
scientist, storyteller, philosopher computingthehumanexperience.com
There is clear and present evidence that all agentic systems of a certain complexity may yield erratic and undesirable behavior. Consider, for example, the election of our current president.
In my work, as in commercial software engineering, we also recognize the presence of something we call “this AI is making shit up again”.
In Victorian times, there was the penny dreadful: mass-produced sensational stories of questionable quality with high shock value, all sold for a penny. I call this new medium of AI slop books "prompt pulp".
Pfff. You young whippersnappers and your fancy machines. I run Kimi K3 locally, the old fashioned way: with pencil and paper, one matrix multiplication at a time.
I already have two books I need to finish writing AND a documentary to produce but if I had the time and space, the book I'd want to write next is "AI Design Patterns".
In the fullness of time, LLMs will eventually become commodities and their price will be a race to the bottom.
How could consciousness know its radiance without passing through the appearance of night? — Russell Schwaber, The Privilege of Forgetting.
Looking to get/find/buy an NVIDIA H100 GPU chip. Not the entire system, not even a functional one, just the packaged chip. Anyone have one that's already end of life?
Looking forward to the release of Elon’s latest project, The Slopyssey.
Why do I host all my git repos locally? Because I don’t want an AI scraping my work.
From a multi-day project with Claude when chasing down a complex software/hardware problem:
I say this with all due respect but simultaneously with a rage that can unmake stars, a software-intensive system that “exhibits all the cognitive capabilities that the brain has” is not “a few short years away”. x.com/demishassabi...
Demis Hassabis (@demishassabis) on X
https://t.co/PTeDiv1b6L
x.com
The world of computing is and will remain far larger, more impactful, and far more vibrant than the narrow subdomain of genAI.
I never expected that my study of Kalman filters in my astrophysics courses at USAFA in the 70s would inform my work on predictive world models in the context of neuro-symbolic systems.
The code is the truth, but it is not the whole truth. A repository is the truth, but it is not the architecture: it represents only the development view of an architecture.
I find the acceptance and defense of mediocrity to be deeply depressing. This is true not just in the context of large language models, but particularly in the case of certain scholars and politicians.
What is the AI equivalent to the Glasgow Coma Scale and the AVPU scale? Asking for a friend.
The brain is a dynamic complex physical system. Every effectively realizable physical system is computable (the Church-Turing thesis). Ergo, the brain is computable. QED
The human brain runs on about 20 watts of power; certain of the frontier large language models require a few gigawatts just to train them and then multiple gigawatts to serve them to their clients.
In my long career, I’ve had the opportunity to work with on the order of two dozen billionaires. A couple of them even offered to hire me for embarrassingly large piles of money. I was flattered, but I turned them both down.
I am one of those whose research points to the conclusion that the mind is computable.