No legacy bug goes unpunished
I was kind of doing a victory lap this morning as I was reading about the PHP vulnerabilities that are causing massive patching across the world's web...
Architectural coherence is at risk
Anthropic released an engineering note this week detailing a migration from git to a source control system they call Scion. Rather than paraphrase, I will...
Why I Worry
Not an AI safety topic in this post. For a while now (at least a year, probably longer) I've been worrying about the financial aspect of the "AI boom." I...
Do Models Reason
I ran across Thore Graepel's article in the MIT Technology Review Don't be fooled, LLM's don't reason, which sounded deliberately provocative. It's worth...
Heterogeneous Agentic Workflows
A kind of comical thing happened yesterday while I was burning a third of a month's AI credits in 3 hours. I had Anthropic's most powerful reasoning model...
SMCC Initial Implementation
It happened again. I came to the end of the month, and I had not (nearly) used up all of my GitHub Copilot "AI Credits", which put me in the luxurious place...
nvidia Open Agent Safety Platform and IFC
The "models and systems" posting should be considered a very helpful pre-read if you have not already read it. I read the various announcements around the...
Formalization, DAGs and Agentic Coding
By now, it's old news that after consuming roughly six billion output tokens, producing 13 million lines of Lean and proving more than 30,000 theorems over...
Linguistic Illegibility
I read with great interest James Mickens' paper The Implications of Linguistic Illegibility for LLM Security. It makes some important points that correlate...
Models and Systems
There has been quite a lot of "chatter" lately about the end of humanity at the "hands" of AI. And much of the commentary has been from a curious bunch, the...
Experiments...
To view in a web browser: https://buttondown.com/eho/archive/experiments First, a quick foray into "endless credits" and Fable 5 before I dive into my first...
Collective Defense
Thanks Phillip, for alerting me to the recent "collective defense" initiative led by OpenAI and signed by, well, just about everyone who matters. A call for...
Open vs. Closed Weights
I saw yet another article (link unnecessary) where Chinese labs (Z.AI in this case) criticize American frontier labs' use of closed weight models as...
DPO and Agent Alignment
I was reconstructing my list of "seminal AI papers" which I fumbled and lost when I exited the corporate firewall. Turns out you can't share an Edge...
Alignment and Reward Hacking (OpenAI/HuggingFace)
I couldn't help myself, during my "weekly AI reading", links for the METR and OpenAI reports on this incident appeared in my feed. So, I read them almost in...
The Genie Problem
Raghavan and Shneier wrote an article in IEEE Spectrum about how we have created genies. AI Agent Benchmarks Need to Measure User Intent - IEEE Spectrum I’m...
AI Inference Server Bring-Up
Our return from our annual “local vacation” was greeted by a Category 1 hurricane, which delayed the bring-up of the new inference server by a few days. We...
3060 eGPU setup
A short one today due to being on a real “away from home” vacation. I finally set up the 3060 in the Sonnet eGPU enclosure. This is an enclosure with power...
Local AI inference build
As I was preparing to leave the enterprise Github Copilot unlimited token all-you-can-eat buffet, I started to put a keen eye on token usage. In the early...
Perception
It’s probably clear from the previous posting on “holding it wrong” that I hadn’t seen the Microsoft Security announcement around MAI-Flash-1-Cyber...