Alex Lindquist
@alxfed.bsky.social
AI Dialogue Facilitator
Ph.D. in Physics and Mathematics
Data Scientist since 1978.
Chicago, IL, since 2003, US Citizen.
"Πάντων χρημάτων μέτρον ἄνθρωπος, τῶν μὲν ὄντων ὡς ἔστι, τῶν δὲ οὐκ ὄντων ὡς οὐκ ἔστι."
In the past Twitter - @alxfed
Ph.D. in Physics and Mathematics
Data Scientist since 1978.
Chicago, IL, since 2003, US Citizen.
"Πάντων χρημάτων μέτρον ἄνθρωπος, τῶν μὲν ὄντων ὡς ἔστι, τῶν δὲ οὐκ ὄντων ὡς οὐκ ἔστι."
In the past Twitter - @alxfed
Pinned
GitHub - alxfed/butterfly: Examples of `blue-yonder` package usage
Examples of `blue-yonder` package usage. Contribute to alxfed/butterfly development by creating an account on GitHub.
github.com
This is a template repository with examples of `blue_yonder` package usage. Click the "Use this template" button at the top of this page and create a copy of this repo in your account where you will be able to use and modify it the way you want.
Tinker looks good enough. Let's see what we are having here...🙂
tinker-docs.thinkingmachines.ai
tinker-docs.thinkingmachines.ai
Tinker: a training API for researchers and developers – Tinker API
Tinker API: A powerful and simple API for training large language models
tinker-docs.thinkingmachines.ai
November 13, 2025 at 2:18 PM
Tinker looks good enough. Let's see what we are having here...🙂
tinker-docs.thinkingmachines.ai
tinker-docs.thinkingmachines.ai
"Every single one must have a share. The fact is, there's no way societies could exist at all if only a few people possessed a sense of right and wrong, the way it is with those other skills. In fact, make it a rule, on my authority, that …
November 13, 2025 at 12:27 PM
"Every single one must have a share. The fact is, there's no way societies could exist at all if only a few people possessed a sense of right and wrong, the way it is with those other skills. In fact, make it a rule, on my authority, that …
Introducing Nested Learning: A new ML paradigm for continual learning
research.google
November 8, 2025 at 7:30 PM
The seal of the Commonwealth of Virginia
November 6, 2025 at 8:31 PM
The seal of the Commonwealth of Virginia
This is a template repository with examples of `blue_yonder` package usage. Click the "Use this template" button at the top of this page and create a copy of this repo in your account where you will be able to use and modify it the way you want.
GitHub - alxfed/butterfly: Examples of `blue-yonder` package usage
Examples of `blue-yonder` package usage. Contribute to alxfed/butterfly development by creating an account on GitHub.
github.com
October 28, 2025 at 11:05 PM
This is a template repository with examples of `blue_yonder` package usage. Click the "Use this template" button at the top of this page and create a copy of this repo in your account where you will be able to use and modify it the way you want.
Tekna is on a "Children's Crusade", which will end up in the same way as the historically first one - the children were sold (and these will be sold too) into slavery.
September 21, 2025 at 1:52 PM
Tekna is on a "Children's Crusade", which will end up in the same way as the historically first one - the children were sold (and these will be sold too) into slavery.
The history of the XX-th century developed in a way that physicists were raised and matured in a World where their craft could 'destroy it all'; they are used to thinking in these terms and are the people least surprised by this 'novelty' Today. Here's one of us: michaelnotebook.com/xriskbrief/
ASI existential risk: reconsidering alignment as a goal
michaelnotebook.com
September 7, 2025 at 2:18 PM
The history of the XX-th century developed in a way that physicists were raised and matured in a World where their craft could 'destroy it all'; they are used to thinking in these terms and are the people least surprised by this 'novelty' Today. Here's one of us: michaelnotebook.com/xriskbrief/
Re-reading Nabokov's "Bend Sinister"; "padograph" is a very good description of a chatbot / Language Model. 🙂
August 27, 2025 at 12:31 PM
Re-reading Nabokov's "Bend Sinister"; "padograph" is a very good description of a chatbot / Language Model. 🙂
"...the whole succession of men, in the course of so many centuries, should be regarded as the same man who exists always and learns continually." (Pascal)
August 14, 2025 at 1:36 PM
"...the whole succession of men, in the course of so many centuries, should be regarded as the same man who exists always and learns continually." (Pascal)
Are Language Models seeking the Truth, Thingking-Machine?
August 8, 2025 at 6:47 PM
Are Language Models seeking the Truth, Thingking-Machine?
Here's a brilliant experiment.
arxiv.org/abs/2507.06952
arxiv.org/abs/2507.06952
July 14, 2025 at 1:07 AM
Here's a brilliant experiment.
arxiv.org/abs/2507.06952
arxiv.org/abs/2507.06952
Fun read. Turned out to be way-way worse because of the money involved.
As We May Think
“Consider a future device ... in which an individual stores all his books, records, and communications, and which is mechanized so that it may be consulted with exceeding speed and flexibility. It is ...
www.theatlantic.com
July 9, 2025 at 10:50 AM
Fun read. Turned out to be way-way worse because of the money involved.
‘...things spoken at random and expressed in the first words that come to mind’
—Plato, “Complete Works”
—Plato, “Complete Works”
July 6, 2025 at 10:39 AM
‘...things spoken at random and expressed in the first words that come to mind’
—Plato, “Complete Works”
—Plato, “Complete Works”
"...and never the twain shall meet."
June 30, 2025 at 11:26 AM
"...and never the twain shall meet."
This! is the real beginning of the demise:
"We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors."
Censorship by Tekna on a Humankind scale. No hesitation. 'We' will.
"We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors."
Censorship by Tekna on a Humankind scale. No hesitation. 'We' will.
June 21, 2025 at 10:44 AM
This! is the real beginning of the demise:
"We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors."
Censorship by Tekna on a Humankind scale. No hesitation. 'We' will.
"We will use Grok 3.5 (maybe we should call it 4), which has advanced reasoning, to rewrite the entire corpus of human knowledge, adding missing information and deleting errors."
Censorship by Tekna on a Humankind scale. No hesitation. 'We' will.
Much ado about... nobody knows what.
Why AI Projects Fail and How They Can Succeed
By some estimates, more than 80 percent of AI projects fail. That's twice the rate of failure of information technology projects that do not involve AI. Interviews with data scientists and engineers i...
www.rand.org
June 20, 2025 at 5:41 PM
Much ado about... nobody knows what.
The Gemini code assistant didn't last long... They completely screwed it up with this update. It only shows a part of code in JetBrains IDEs.
June 10, 2025 at 7:37 PM
The Gemini code assistant didn't last long... They completely screwed it up with this update. It only shows a part of code in JetBrains IDEs.
"Like painted kites
Those days and nights they went flyin' by..."
Those days and nights they went flyin' by..."
May 24, 2025 at 7:57 PM
"Like painted kites
Those days and nights they went flyin' by..."
Those days and nights they went flyin' by..."
I realized that these enormous crowds of slave-coders proudly calling themselves 'software engineers' are absolutely the 'proletarians' of the XXI-th century in that old Marxian sense. They own nothing (including their 'code') and are fully dependent on the 'job givers' to support their existence.
May 7, 2025 at 12:43 PM
I realized that these enormous crowds of slave-coders proudly calling themselves 'software engineers' are absolutely the 'proletarians' of the XXI-th century in that old Marxian sense. They own nothing (including their 'code') and are fully dependent on the 'job givers' to support their existence.
Google has made a _very_ big mistake by assigning 'system_instructions' for the new 'thinking' models to the role 'user'. Now they are just coming from the user as a first part of the text.
For models of other companies system instruction is an absolute 'voice of God', that they can not disobey.
For models of other companies system instruction is an absolute 'voice of God', that they can not disobey.
April 23, 2025 at 8:39 PM
Google has made a _very_ big mistake by assigning 'system_instructions' for the new 'thinking' models to the role 'user'. Now they are just coming from the user as a first part of the text.
For models of other companies system instruction is an absolute 'voice of God', that they can not disobey.
For models of other companies system instruction is an absolute 'voice of God', that they can not disobey.
grok.com/workspace is pretty usable.
Grok
Grok is a free AI assistant designed by xAI to maximize truth and objectivity. Grok offers real-time search, image generation, trend analysis, and more.
grok.com
April 16, 2025 at 2:42 PM
grok.com/workspace is pretty usable.
It is absolutely symbolic that the 'business-process' in 'NotebookLM' (notebooklm.google.com) doesn't start with a human writing down her or his thought, it starts with an 'upload source' or 'paste copied text'.
Tekna don't have thought of their own, they endlessly 'rewrite', what other people wrote
Tekna don't have thought of their own, they endlessly 'rewrite', what other people wrote
April 12, 2025 at 12:27 PM
It is absolutely symbolic that the 'business-process' in 'NotebookLM' (notebooklm.google.com) doesn't start with a human writing down her or his thought, it starts with an 'upload source' or 'paste copied text'.
Tekna don't have thought of their own, they endlessly 'rewrite', what other people wrote
Tekna don't have thought of their own, they endlessly 'rewrite', what other people wrote
This is an important idea.
arxiv.org/abs/2504.019...
arxiv.org/abs/2504.019...
Is the Reversal Curse a Binding Problem? Uncovering Limitations of Transformers from a Basic Generalization Failure
Despite their impressive capabilities, LLMs exhibit a basic generalization failure known as the Reversal Curse, where they struggle to learn reversible factual associations. Understanding why this occ...
arxiv.org
April 9, 2025 at 10:38 AM
This is an important idea.
arxiv.org/abs/2504.019...
arxiv.org/abs/2504.019...