Carlos Barth
@cbarth.me
Post-doc researcher in Philosophy of Cognition and AI. Dev and Software Libre (FOSS) enthusiast (specially infosec). (EN/PT)
"This is a fundamentally different kind of work. Creating is energizing. Reviewing is draining. (..) Generative work gives you flow states. Evaluative work gives you decision fatigue."
siddhantkhare.com/writing/ai-f...
siddhantkhare.com/writing/ai-f...
AI fatigue is real and nobody talks about it | Siddhant Khare
You're using AI to be more productive. So why are you more exhausted than ever? The paradox every engineer needs to confront.
siddhantkhare.com
February 8, 2026 at 4:36 PM
"This is a fundamentally different kind of work. Creating is energizing. Reviewing is draining. (..) Generative work gives you flow states. Evaluative work gives you decision fatigue."
siddhantkhare.com/writing/ai-f...
siddhantkhare.com/writing/ai-f...
Their argument is not sound, but it is informative paying attention to what they consider "evidence" for AGI. A nice instance of a problem that seems peculiar to AI: it tries to define both its target phenomenon and how well it is doing towards it.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Does AI already have human-level intelligence? The evidence is clear
The vision of human-level machine intelligence laid out by Alan Turing in the 1950s is now a reality. Eyes unclouded by dread or hype will help us to prepare for what comes next.
www.nature.com
February 4, 2026 at 1:06 PM
Their argument is not sound, but it is informative paying attention to what they consider "evidence" for AGI. A nice instance of a problem that seems peculiar to AI: it tries to define both its target phenomenon and how well it is doing towards it.
www.nature.com/articles/d41...
www.nature.com/articles/d41...
Vibe coding is somewhat like trying to learn a language exclusively through duolingo-like apps. You think you're getting somewhere, until you realize you're not.
atmoio.substack.com/p/after-two-...
atmoio.substack.com/p/after-two-...
After two years of vibecoding, I'm back to writing by hand
Agents write units of changes that look good in isolation. They are consistent with themselves and your prompt. But respect for the whole, there is not.
atmoio.substack.com
January 27, 2026 at 8:34 PM
Vibe coding is somewhat like trying to learn a language exclusively through duolingo-like apps. You think you're getting somewhere, until you realize you're not.
atmoio.substack.com/p/after-two-...
atmoio.substack.com/p/after-two-...
Quick thought: I don't see many students or researchers realizing that their (unethical) reliance on AI can haunt them in the far future. Even if no one notices that your text is not really yours today, someone (or a better AI detection tool) can do so in the future.
January 18, 2026 at 10:28 PM
Quick thought: I don't see many students or researchers realizing that their (unethical) reliance on AI can haunt them in the far future. Even if no one notices that your text is not really yours today, someone (or a better AI detection tool) can do so in the future.
Reposted by Carlos Barth
Martin Peterson's creative response to being banned from teaching Plato (shared with his permission).
January 8, 2026 at 5:38 PM
Martin Peterson's creative response to being banned from teaching Plato (shared with his permission).
Considering the current state of affairs in the world, I'm officially switching from "good morning" to "good luck". Good luck everyone.
January 8, 2026 at 4:25 PM
Considering the current state of affairs in the world, I'm officially switching from "good morning" to "good luck". Good luck everyone.
Mujica foi um raríssimo exemplo de alguém capaz de viver inteiramente segundo os princípios em que acreditava e pelos quais lutava.
Ainda que nunca pare de nos inspirar, o mundo de hoje é pior que o de ontem.
Ainda que nunca pare de nos inspirar, o mundo de hoje é pior que o de ontem.
May 13, 2025 at 9:49 PM
Mujica foi um raríssimo exemplo de alguém capaz de viver inteiramente segundo os princípios em que acreditava e pelos quais lutava.
Ainda que nunca pare de nos inspirar, o mundo de hoje é pior que o de ontem.
Ainda que nunca pare de nos inspirar, o mundo de hoje é pior que o de ontem.
Consent is not consent when it's not voluntary.
May 5, 2025 at 1:20 PM
Consent is not consent when it's not voluntary.
This bothers me sometimes: what, in our zeitgeist, accounts for people's lack of care about data privacy?
Whatever it is, it seems immune to knowledge.
People can learn it's about power, or the role of their data in business models, and still refuse to make any privacy-increasing move.
Whatever it is, it seems immune to knowledge.
People can learn it's about power, or the role of their data in business models, and still refuse to make any privacy-increasing move.
April 2, 2025 at 8:10 PM
This bothers me sometimes: what, in our zeitgeist, accounts for people's lack of care about data privacy?
Whatever it is, it seems immune to knowledge.
People can learn it's about power, or the role of their data in business models, and still refuse to make any privacy-increasing move.
Whatever it is, it seems immune to knowledge.
People can learn it's about power, or the role of their data in business models, and still refuse to make any privacy-increasing move.
From the same kind of guy that in the 1990's used to tell that Internet will render big techs useless.
March 28, 2025 at 6:04 PM
From the same kind of guy that in the 1990's used to tell that Internet will render big techs useless.
Not so far from the most likely outcome.
March 28, 2025 at 6:02 PM
Not so far from the most likely outcome.
Tamo junto, Alfredo!
February 19, 2025 at 11:56 AM
Tamo junto, Alfredo!
Nice point!
My take on the main lesson from the point of view of #AIEthics regarding DeepSeek: if we had imposed constraints for ethical reasons on American and European models, maybe we would've gotten to a more effective model before a Chinese company. A short thread 1/ 👇
January 27, 2025 at 11:19 PM
Nice point!
The more I use social networks, the more I miss good old mail lists.
No bigtech-dependence and heavily moderated even on broad topics (no one took that as threat to freedom of speech).
Saying dumb stuff was socially costly. As a result, most would think twice and do their homework before writing.
No bigtech-dependence and heavily moderated even on broad topics (no one took that as threat to freedom of speech).
Saying dumb stuff was socially costly. As a result, most would think twice and do their homework before writing.
January 24, 2025 at 1:26 PM
The more I use social networks, the more I miss good old mail lists.
No bigtech-dependence and heavily moderated even on broad topics (no one took that as threat to freedom of speech).
Saying dumb stuff was socially costly. As a result, most would think twice and do their homework before writing.
No bigtech-dependence and heavily moderated even on broad topics (no one took that as threat to freedom of speech).
Saying dumb stuff was socially costly. As a result, most would think twice and do their homework before writing.
Precisamos normalizar a vontade de gritar quando pedem nome completo, RG, CPF, endereço, CEP, e-mail e celular só pra marcar um cabeleireiro.
January 21, 2025 at 7:19 PM
Precisamos normalizar a vontade de gritar quando pedem nome completo, RG, CPF, endereço, CEP, e-mail e celular só pra marcar um cabeleireiro.
Bad news: due to unforeseen issues, the CLEA Invites with talks from William Ramsey and Ian Robertson had will no longer happen tomorrow.
Good news: it will take place on December 18, so no one has to wait that much.
If you're not registered already, consider this your new invitation to join us!
Good news: it will take place on December 18, so no one has to wait that much.
If you're not registered already, consider this your new invitation to join us!
December 12, 2024 at 10:04 PM
Bad news: due to unforeseen issues, the CLEA Invites with talks from William Ramsey and Ian Robertson had will no longer happen tomorrow.
Good news: it will take place on December 18, so no one has to wait that much.
If you're not registered already, consider this your new invitation to join us!
Good news: it will take place on December 18, so no one has to wait that much.
If you're not registered already, consider this your new invitation to join us!
The last CLEA Invites of the year is almost here!
Register to get the meeting link and join us for a nice and friendly discussion on the hard problem of content and related foundational issues within philosophy of cognition.
Hope to see you there!
clea.group/register
Register to get the meeting link and join us for a nice and friendly discussion on the hard problem of content and related foundational issues within philosophy of cognition.
Hope to see you there!
clea.group/register
December 11, 2024 at 8:32 PM
The last CLEA Invites of the year is almost here!
Register to get the meeting link and join us for a nice and friendly discussion on the hard problem of content and related foundational issues within philosophy of cognition.
Hope to see you there!
clea.group/register
Register to get the meeting link and join us for a nice and friendly discussion on the hard problem of content and related foundational issues within philosophy of cognition.
Hope to see you there!
clea.group/register
"How can I be more Neural Network today?"
December 11, 2024 at 8:30 PM
"How can I be more Neural Network today?"
Reposted by Carlos Barth
thinking of calling this "The Illusion Illusion"
(more examples below)
(more examples below)
December 1, 2024 at 2:33 PM
thinking of calling this "The Illusion Illusion"
(more examples below)
(more examples below)
Reposted by Carlos Barth
"None of us thought, when we invented social media, that we would become a threat to democracy." Not a great excuse. 1. It could've been designed better, with #AIEthics and #privacy in mind. 2. We could've improved it years ago, after learning from experience
www.businessinsider.com/eric-schmidt...
www.businessinsider.com/eric-schmidt...
Ex-Google CEO Eric Schmidt says AI will 'shape' identity and that 'normal people' are not ready for it
Former Google CEO Eric Schmidt said in the future AI could be a child's best friend, shaping their identity and worldview.
www.businessinsider.com
November 24, 2024 at 6:14 PM
"None of us thought, when we invented social media, that we would become a threat to democracy." Not a great excuse. 1. It could've been designed better, with #AIEthics and #privacy in mind. 2. We could've improved it years ago, after learning from experience
www.businessinsider.com/eric-schmidt...
www.businessinsider.com/eric-schmidt...
The next meeting of the CLEA Research Group is here!
This time we'll discuss problems concerning the contents of mental representations through the work of William Ramsey and Ian Robertson.
Join us!
This time we'll discuss problems concerning the contents of mental representations through the work of William Ramsey and Ian Robertson.
Join us!
November 24, 2024 at 11:38 AM
The next meeting of the CLEA Research Group is here!
This time we'll discuss problems concerning the contents of mental representations through the work of William Ramsey and Ian Robertson.
Join us!
This time we'll discuss problems concerning the contents of mental representations through the work of William Ramsey and Ian Robertson.
Join us!
Reposted by Carlos Barth
Signatures of behaviour in visual representations; affordances in the brain?
www.nature.com/articles/s41...
www.nature.com/articles/s41...
Distributed representations of behaviour-derived object dimensions in the human visual system - Nature Human Behaviour
Contier et al. show that dimensions are superior to categories at predicting brain responses to visual objects.
www.nature.com
September 11, 2024 at 5:55 AM
Signatures of behaviour in visual representations; affordances in the brain?
www.nature.com/articles/s41...
www.nature.com/articles/s41...