Information from Documents
@docs2info.bsky.social
Read documents. Not too many. Mostly PDFs.
Our parents read and wrote documents. Having the world's knowledge an internet connection away doesn't change this.
Building modest data tools for independents.
https://docs2info.com
Our parents read and wrote documents. Having the world's knowledge an internet connection away doesn't change this.
Building modest data tools for independents.
https://docs2info.com
The first rule of AI take-over. Always scan the survivors' brains.
There it is: doomsday if we don't slow down.
This isn't forecast. It's propaganda.
This isn't forecast. It's propaganda.
April 5, 2025 at 1:58 AM
The first rule of AI take-over. Always scan the survivors' brains.
Reposted by Information from Documents
You’ve probably heard about how AI/LLMs can solve Math Olympiad problems ( deepmind.google/discover/blo... ).
So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released.
The result: They all sucked!
So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released.
The result: They all sucked!
March 31, 2025 at 8:33 PM
You’ve probably heard about how AI/LLMs can solve Math Olympiad problems ( deepmind.google/discover/blo... ).
So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released.
The result: They all sucked!
So naturally, some people put it to the test — hours after the 2025 US Math Olympiad problems were released.
The result: They all sucked!
Welcome to the duck-side of vendor benchmarking.
March 29, 2025 at 2:11 AM
Welcome to the duck-side of vendor benchmarking.
Today's software developers, my peers and mentors, look back at past generations with pity. What could people from these old professions achieved if they had learned to code? Web pages, FAANG jobs, mastery of the Agile ceremonies, untold potential, ... x.com/docs2info/st...
Information from Documents on X: "@bobvanluijt Or you could just write out your thoughts long-hand and meet every few weeks. That worked well enough for Darwin, Curie, Einstein and their ilk. Their modern equivalents, scrum masters, social media influencers and vibe coders, might find this kind of communication limiting." / X
@bobvanluijt Or you could just write out your thoughts long-hand and meet every few weeks. That worked well enough for Darwin, Curie, Einstein and their ilk. Their modern equivalents, scrum masters, social media influencers and vibe coders, might find this kind of communication limiting.
x.com
March 28, 2025 at 10:28 PM
Today's software developers, my peers and mentors, look back at past generations with pity. What could people from these old professions achieved if they had learned to code? Web pages, FAANG jobs, mastery of the Agile ceremonies, untold potential, ... x.com/docs2info/st...
This is where we are in 2025. x.com/HanchungLee/...
Han on X: "still vibe coding or rizz coding? time to jizz code. https://t.co/aXPzOi1aAF" / X
still vibe coding or rizz coding? time to jizz code. https://t.co/aXPzOi1aAF
x.com
March 26, 2025 at 9:40 AM
This is where we are in 2025. x.com/HanchungLee/...
Reposted by Information from Documents
Trying something new:
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
November 20, 2024 at 5:09 PM
Trying something new:
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
A 🧵 on a topic I find many students struggle with: "why do their 📊 look more professional than my 📊?"
It's *lots* of tiny decisions that aren't the defaults in many libraries, so let's break down 1 simple graph by @jburnmurdoch.bsky.social
🔗 www.ft.com/content/73a1...
Reposted by Information from Documents
Open source is draining.
One's todo-list is open to world.
People seldom realize the cost of what they get for free.
Unpleasant comments do happen.
Cost of maintenance is not understood.
opensource.com/article/17/2...
One's todo-list is open to world.
People seldom realize the cost of what they get for free.
Unpleasant comments do happen.
Cost of maintenance is not understood.
opensource.com/article/17/2...
March 22, 2025 at 3:53 PM
Open source is draining.
One's todo-list is open to world.
People seldom realize the cost of what they get for free.
Unpleasant comments do happen.
Cost of maintenance is not understood.
opensource.com/article/17/2...
One's todo-list is open to world.
People seldom realize the cost of what they get for free.
Unpleasant comments do happen.
Cost of maintenance is not understood.
opensource.com/article/17/2...
A Google executive tells programmers to use Google products and not worry about their job security. www.linkedin.com/feed/update/...
Vibe Coding is a Dangerous Fantasy | Karthik Ramgopal | 31 comments
A lot of what this article says is true. The coding AI assistants of today can generate a lot of code, but the generated code is often ridden with (subtle)… | 31 comments on LinkedIn
www.linkedin.com
March 23, 2025 at 10:53 PM
A Google executive tells programmers to use Google products and not worry about their job security. www.linkedin.com/feed/update/...
The resistance has started. arstechnica.com/ai/2025/03/c...
Cloudflare turns AI against itself with endless maze of irrelevant facts
New approach punishes AI companies that ignore “no crawl” directives.
arstechnica.com
March 23, 2025 at 6:44 AM
The resistance has started. arstechnica.com/ai/2025/03/c...
Reposted by Information from Documents
The promise of AI in education is it will save educators time. That's rubbish. It's sapping all our time already in working groups, committees, workshops, and marking anxiety. It's the most time-sapping, exhausting and frankly boring thing I've ever encountered in HE - and it's in my research area.
March 21, 2025 at 11:51 PM
The promise of AI in education is it will save educators time. That's rubbish. It's sapping all our time already in working groups, committees, workshops, and marking anxiety. It's the most time-sapping, exhausting and frankly boring thing I've ever encountered in HE - and it's in my research area.
Christmas Eve 2024. George Bailey fights AI-driven job loss, facing ruin from OpenAI. Clarence Odbody shows him a bleak world without his advocacy. Inspired, George leads his community to develop ethical tech and push for fairness, proving humanity’s resilience can ensure technology benefits all.
December 27, 2024 at 9:54 PM
Christmas Eve 2024. George Bailey fights AI-driven job loss, facing ruin from OpenAI. Clarence Odbody shows him a bleak world without his advocacy. Inspired, George leads his community to develop ethical tech and push for fairness, proving humanity’s resilience can ensure technology benefits all.
Reposted by Information from Documents
Yet again we see a situation in the wild where a gzipped csv would be infinitely more helpful than a restful http api. But web app devs gotta web app dev.
My personal project is to take congress.gov congressional bills data and load it into bigquery tables but I think I'd have to do it 250 records per request, 5000 requests per hour until i have the whole population. Pretty sure I can figure this all out & just iterate but whatdahell.
December 24, 2024 at 3:38 AM
Yet again we see a situation in the wild where a gzipped csv would be infinitely more helpful than a restful http api. But web app devs gotta web app dev.
Reposted by Information from Documents
Merry Slopmas www.404media.co/email/d77d05...
Merry Slopmas!
AI-generated Christmas classics that dwell in the uncanny valley are giving listeners the creeps.
www.404media.co
December 25, 2024 at 9:33 PM
Merry Slopmas www.404media.co/email/d77d05...
December 25, 2024 at 9:39 PM
Reposted by Information from Documents
Fancy GenAI stuff like GPT 4 is too big, slow, private, and expensive for many jobs. Consider that the original GPT-1 was 117m params. Llama 3.1, by contrast, has up to 405 billion params! 😲
These models are slow, expensive, and *not yours to control*.
These models are slow, expensive, and *not yours to control*.
December 19, 2024 at 4:45 PM
Fancy GenAI stuff like GPT 4 is too big, slow, private, and expensive for many jobs. Consider that the original GPT-1 was 117m params. Llama 3.1, by contrast, has up to 405 billion params! 😲
These models are slow, expensive, and *not yours to control*.
These models are slow, expensive, and *not yours to control*.
Reposted by Information from Documents
"pre training as we know it will end (because we will run out of data)" is, in other words, "learning to complete partial observations is not sufficient to get to intelligence". i think this was kinda obvious to many, but maybe noteworthy that a true scale-believer said it.
December 14, 2024 at 12:32 PM
"pre training as we know it will end (because we will run out of data)" is, in other words, "learning to complete partial observations is not sufficient to get to intelligence". i think this was kinda obvious to many, but maybe noteworthy that a true scale-believer said it.
Reposted by Information from Documents
I wrote something about Google's Willow for The Marginalia Review of Books,and what it means (and what it doesn't).
(Ignore a couple of formatting errors: 3x3 etc, and 10^25 - we'll fix them.)
www.marginaliareviewofbooks.com/post/google-...
(Ignore a couple of formatting errors: 3x3 etc, and 10^25 - we'll fix them.)
www.marginaliareviewofbooks.com/post/google-...
Google's Willow and the Future of Quantum Computing
PHILIP BALL | I’d put money on quantum computing becoming commercially viable and useful way before fusion does...
www.marginaliareviewofbooks.com
December 13, 2024 at 9:55 PM
I wrote something about Google's Willow for The Marginalia Review of Books,and what it means (and what it doesn't).
(Ignore a couple of formatting errors: 3x3 etc, and 10^25 - we'll fix them.)
www.marginaliareviewofbooks.com/post/google-...
(Ignore a couple of formatting errors: 3x3 etc, and 10^25 - we'll fix them.)
www.marginaliareviewofbooks.com/post/google-...
Reposted by Information from Documents
“To write well is to think clearly” - which is to say that by delegating more and more of your writing effort to AI, you are inviting the slow erosion of your own ability to think clearly.
December 13, 2024 at 3:56 AM
“To write well is to think clearly” - which is to say that by delegating more and more of your writing effort to AI, you are inviting the slow erosion of your own ability to think clearly.
Workers who value their own labor will empower themselves with AI while their supervisors make work for others.
"I once asked a PhD student to go through the 100 or so reports on innovation in Australia I had on my shelf to look for cross-references."
www.aumanufacturing.com.au/why-reviews-...
"I once asked a PhD student to go through the 100 or so reports on innovation in Australia I had on my shelf to look for cross-references."
www.aumanufacturing.com.au/why-reviews-...
Why reviews of R&D fail - by Mark Dodgson - Australian Manufacturing Forum
The federal government has announced a strategic examination of R&D, the latest in a series of inquiries into Australia’s faltering innovation system. Here Mark Dodgson looks at why previous reviews h...
www.aumanufacturing.com.au
December 13, 2024 at 4:15 AM
Workers who value their own labor will empower themselves with AI while their supervisors make work for others.
"I once asked a PhD student to go through the 100 or so reports on innovation in Australia I had on my shelf to look for cross-references."
www.aumanufacturing.com.au/why-reviews-...
"I once asked a PhD student to go through the 100 or so reports on innovation in Australia I had on my shelf to look for cross-references."
www.aumanufacturing.com.au/why-reviews-...
People who worry about governments making gifts to favored businesses should take their eyes off Mr Trump and look to Australia.
Government handout like these won't make me read mainstream media. I don't like its dull PR pretending to be news, either directly or echoed on Google and social media.
Government handout like these won't make me read mainstream media. I don't like its dull PR pretending to be news, either directly or echoed on Google and social media.
December 12, 2024 at 11:31 AM
People who worry about governments making gifts to favored businesses should take their eyes off Mr Trump and look to Australia.
Government handout like these won't make me read mainstream media. I don't like its dull PR pretending to be news, either directly or echoed on Google and social media.
Government handout like these won't make me read mainstream media. I don't like its dull PR pretending to be news, either directly or echoed on Google and social media.
Reposted by Information from Documents
The evolution of free will - with Kevin Mitchell youtu.be/QXwRjb7WF7E?... - my talk at the Royal Institution earlier this year 😊
The evolution of free will - with Kevin Mitchell
YouTube video by The Royal Institution
youtu.be
December 12, 2024 at 9:38 AM
The evolution of free will - with Kevin Mitchell youtu.be/QXwRjb7WF7E?... - my talk at the Royal Institution earlier this year 😊
Reposted by Information from Documents
AI hype is making AI researchers forget painfully learned lessons core to the field.
There's an emerging cope that progress in capabilities isn't slowing down—it's just invisible as benchmarks are saturated; vibe checks are useless because models are now superhuman so we can't perceive improvement.
There's an emerging cope that progress in capabilities isn't slowing down—it's just invisible as benchmarks are saturated; vibe checks are useless because models are now superhuman so we can't perceive improvement.
December 11, 2024 at 12:58 PM
AI hype is making AI researchers forget painfully learned lessons core to the field.
There's an emerging cope that progress in capabilities isn't slowing down—it's just invisible as benchmarks are saturated; vibe checks are useless because models are now superhuman so we can't perceive improvement.
There's an emerging cope that progress in capabilities isn't slowing down—it's just invisible as benchmarks are saturated; vibe checks are useless because models are now superhuman so we can't perceive improvement.
Reposted by Information from Documents
NEW: A new lawsuit calls Scale AI 'the sordid underbelly of the generative AI industry,' alleges a vast slate of CA labor law violations and says workers were emotionally disturbed by working on a self-harm prevention AI tool for Meta.
www.inc.com/sam-blum/sca...
www.inc.com/sam-blum/sca...
Lawsuit Calls Scale AI ‘The Sordid Underbelly of the Generative AI Industry’
CEO Alexandr Wang is named in the suit, which cites emotional distress for contractors who worked on self-harm prevention project for Meta.
www.inc.com
December 11, 2024 at 7:14 PM
NEW: A new lawsuit calls Scale AI 'the sordid underbelly of the generative AI industry,' alleges a vast slate of CA labor law violations and says workers were emotionally disturbed by working on a self-harm prevention AI tool for Meta.
www.inc.com/sam-blum/sca...
www.inc.com/sam-blum/sca...