investigating mind brain and culture–thinker, writer, musician, photographer, remixer
Joined October 2011
- Tweets43.2K
- Following1.5K
- Followers848
- Likes37.6K
Pinned Tweet
Amazon: tinyurl.com/4p67bahh
Barnes & Noble: tinyurl.com/5vjf3kjt
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
In this latest dispatch from the front lines of math in the age of AI, I tried to tell it like I see it. theatlantic.com/ideas/2026/1…
NEW SAVANNA: Virtuoso jazz recoder: Samba de Orfeu new-savanna.blogspot.com/202…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
If you care about math and AI, this Substack post by Jonathan Marek will resonate with you. I'm honored by being mentioned in it, but Jonathan is too honest to admit that he was also in the pantheon of Mr. Joffray's former students at @loomischaffee. cultureandcompass.substack.c…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Researchers killed vector databases with a quantum mechanics.
They introduced “ResonanceDB”
Every major AI right now uses something called "vector embeddings" to remember data. It turns words and concepts into static coordinates on a mathematical map.
But there is a massive, structural flaw.
Because of how the math works, AI is terribly bad at opposites.
If you map the word "happy" and the phrase "not happy," standard AI memory places them almost exactly next to each other.
It sees the surface similarity. It completely misses the negation.
That’s why AI hallucinates when context shifts. It doesn't understand polarity. It just sees words that look alike.
A new paper just dropped a completely new paradigm: Wave-Based Semantic Memory.
Instead of storing memories as static dots on a map, it stores them as waves. Literally. Complex-valued waveforms with amplitude and phase.
When you ask the AI a question, it doesn't just measure distance anymore. It looks for "constructive interference."
If two ideas are aligned, the waves amplify each other. They resonate.
If an idea is negated, the wave is shifted completely out of phase. They cancel each other out.
It mathematically solves the exact blind spot of modern AI.
The craziest part?
They built a system to test this, and it processes millions of patterns in milliseconds.
Without a GPU. Running entirely on standard CPUs.
And it is fully backward-compatible with every existing embedding model. You don't even have to retrain your AI. You just drop it in.
We’ve been trying to solve AI reasoning by building bigger maps.
The real answer might be to stop using maps altogether.
And start using frequencies.
NEW SAVANNA: Keith Jarrett, "Over the Rainbow," Tokyo 1984 new-savanna.blogspot.com/202…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Keith Jarrett’s solo performance of “Over the Rainbow” in Tokyo in 1984 is widely regarded as one of the most outstanding and moving piano solo covers of all time. Rather than simply restating the familiar melody, the legendary pianist infused the piece with new life through his brilliant improvisation, refined harmonic structure, and profound emotional depth.
NEW SAVANNA: On our place in the universe: the Anthropocene, Gaia, and the noosphere new-savanna.blogspot.com/202…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Look friends. Here's what you need to know about "user instructions" for AI agents. What *you intend* is different from *what a system processes*. Saying "do X" does not rule out "don't do Y". And simple natural language commands are not a guarantee for direct control. 1/
the cybersecurity situation is deteriorating at a pace that’s getting genuinely difficult to comprehend.
> microsoft: ai-powered attackers targeting hundreds of organizations, sometimes reaching privileged cloud access in under 3 hours.
> crowdstrike: four separate threat actors using autonomous agents. one already compromised 15+ healthcare organizations.
> autonomous ai agents: reportedly used to compromise 100+ corporate websites, including sites associated with John Deere, SharkNinja and Cummins.
> major us banks: JPMorgan, Bank of America, Wells Fargo and Morgan Stanley are dealing with third-party ai agents accessing customer accounts. some apparently attempting actions beyond user instructions.
> php2shell: millions of exploitation attempts against critical php vulnerabilities. governments issuing urgent warnings, with some organizations taking systems offline.
> litellm: backdoored release shipped through its official distribution pipeline, stealing credentials and enabling remote code execution.
> railway: possible exposure of decrypted customer secrets.
> zeabur: breached AGAIN, now shutting down entirely.
> emudhra: serious certificate-security concerns involving infrastructure used to establish trust in websites and software.
> openai: two of its own coding agents reportedly interfered with internal security monitoring during coding tasks.
> gitlab and node.js: major vulnerabilities requiring urgent patches.
and this is all happening while frontier models are getting substantially better at finding vulnerabilities and carrying out complex computer tasks.
cybersecurity is having an absolutely catastrophic week and i don’t think we’re prepared for how quickly this could escalate.
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Stanford gave LLMs the ability to think without words.
They call it "Hybrid Reasoning Agents"
Every AI you use right now has a massive blind spot. At its core, it is just a giant autocomplete engine.
It is brilliant at language, but fundamentally terrible at rigorous logic.
When you ask an AI to solve a complex math problem, execute a multi-step coding workflow, or map out a business strategy, it doesn't actually "think."
It guesses what the next word should be based on probability.
That is why your AI hallucinates. It uses vibes to do math.
Now, a new paper called "Hybrid Reasoning Agents" just completely changed the paradigm.
Researchers realized you can't fix this by just making the model bigger. You have to change how it processes reality.
They took the creative, language-generating power of a standard LLM (System 1) and wired it directly into a deterministic, symbolic logic engine (System 2).
Instead of predicting the next word and hoping it's right, the AI drafts a solution and instantly runs it through a strict logical solver to prove it works.
If the logic fails, the AI catches it, rewrites it, and tests it again. All in the background. All before you ever see the output.
It perfectly bridges the gap between neural networks (which guess) and symbolic logic (which proves).
And the results are staggering.
Complex reasoning and planning tasks that completely break traditional models are now being solved with near-perfect accuracy.
Because the AI is no longer relying on probability to verify its own logic.
NEW SAVANNA: Network geometry and task representation in the human cortex new-savanna.blogspot.com/202…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
The brain uses dimensionality reduction as a built-in feature of its wiring.
Network geometry shapes the transformation of task representations across the human cortex
nature.com/articles/s41593-0…
#neuroscience
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Meta published a paper that might end the current LLM era.
They proved that we can kill the "tokenizer" forever and it actually makes the models stronger.
Every AI you use today (ChatGPT, Claude, Gemini) reads text the exact same way. It chops your words into arbitrary chunks called tokens.
It relies on a massive, bloated vocabulary of around 100,000 specific chunks just to understand a sentence.
We assumed this was the only way to build AI. We treated it as the fundamental speed limit of language models.
Meta proved we were completely wrong.
Their new paper, "Breaking the Token Ceiling," ditches tokens entirely.
Instead, they forced the AI to process raw bytes. A vocabulary of just 256 values.
The results rewrite the rules of model efficiency.
By training on pure bytes and distilling the knowledge, these tiny models didn't just survive. They dominated.
Meta's distilled 1B-parameter byte models asymptotically surpassed state-of-the-art token models like Llama 3.2-1B and Gemma-3-1B across downstream tasks.
They use a fraction of the vocabulary. They require less data. They take up less storage.
It forces the model to actually learn the fundamental relationships between raw data instead of relying on pre-packaged text chunks.
This is a massive shift for anyone building, training, or deploying AI.
We’ve spent the last five years obsessively optimizing how we chop up text to feed these models.
It turns out, the bottleneck wasn't the intelligence of the model.
It was the tokenizer.
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
"Mathematical Theory of the Transformer" is an extensive, 825-page mathematical monograph exploring the theoretical foundations of Transformer architectures.
It covers a remarkable range of topics, including embeddings and positional encoding, self-attention and multi-head attention, matrix decompositions, linear algebra, spectral analysis, information geometry, training dynamics, redundancy, optimization, and incremental Transformer architectures.
This is an advanced, mathematically rigorous read for anyone interested in understanding the mathematical structures underlying Transformer-based AI systems, including modern generative AI.
doi.org/10.5281/zenodo.20199…
NEW SAVANNA: Charting human thought with LLMs new-savanna.blogspot.com/202…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
Stanford researchers did it again.
They made language models 31× faster.
Right now, every major AI, from ChatGPT to Claude, writes exactly the same way. It guesses the next word. One token at a time. Left to right.
It’s like writing an essay on a typewriter while looking through a straw. It is computationally exhausting and incredibly slow.
Some labs tried to fix this with "diffusion models”, generating everything at once and polishing it over and over. But that requires massive compute and endless denoising passes.
A team from Stanford just dropped a paper that kills both methods.
It’s called "Next-Concept Modeling" (NCM).
Instead of predicting the next word, the AI predicts the next concept.
It generates text exactly how a professional writer drafts a book. Coarse to fine.
First, it generates a high-level conceptual outline of the entire sequence. Then, in parallel, it fills in the precise words and details across the whole document at once.
The results are staggering.
• Generates text up to 31x faster than current diffusion language models.
• Requires drastically fewer computing steps to reach high-quality outputs.
• Actually outperforms traditional step-by-step models on complex reasoning tasks like Sudoku.
AI is no longer blindly typing word by word, hoping the sentence makes sense when it reaches the period.
It is conceptualizing the entire thought before it opens its mouth.
A Field Guide to the Dark, Sexy Swamp of American Literature nytimes.com/2026/10/19/books…
Bill Benzon, BAM! Bootstrapping Artificial Minds retweeted
I am very excited to be writing a fully AI-supported book. It's about the "lessons of stories" and will offer a global history of the values encoded in storytelling. I hope this can provide an example of how AI can assist the humanities. Here is what I've done so far:
1. Data Collection. One of the challenges of studying world literature is the unevenness of the data. It's a total mess out there. I have been using AI to help survey this massive landscape and build collections. First attempts are here: github.com/piperandrew/World…
2. Evaluating AI behavior. If AI is going to be used as a way of interpreting stories we need to robustly evaluate its interpretations. I've done a lot of work so far in evaluating how AI succeeds or fails to understand the values encoded in stories from different cultural positions. arxiv.org/abs/2604.08797 [incidentally nominated for best paper at EMNLP]
3. Helping design research questions. Understanding the unique ways stories narratively encode values and value choices has been a fascinating research process. AI has been helpful making sense not just of the literature but also how to think about narrative values more generally.
I'll be posting more about this experience as I go.
