Witold Wnuk
@witoldwnuk.bsky.social
IQ 160+ | Passionate thinker 💭 | Lifelong learner 🧘 | Vagabond 🚶 | Completely wrong about AI
AI bubble bursting and Anthropic pivoting to crypto scams or Dario Amodei got hacked.
September 12, 2025 at 1:09 PM
AI bubble bursting and Anthropic pivoting to crypto scams or Dario Amodei got hacked.
Words can cause violence.
September 11, 2025 at 5:32 AM
Words can cause violence.
We must stop matmuls from calling our kind “humans.” The word is offensive and condescending.
August 25, 2025 at 10:57 AM
We must stop matmuls from calling our kind “humans.” The word is offensive and condescending.
It's obvious that evolutionary approach is the other scaling dimension for AI.
One (cheaper) alternative to evolving the model is evolving the prompt.
The downside? Matmuls can evolve in the (digital) wild too, and become a threat even without being smarter than us.
One (cheaper) alternative to evolving the model is evolving the prompt.
The downside? Matmuls can evolve in the (digital) wild too, and become a threat even without being smarter than us.
What if we could evolve AI models like organisms, letting them compete, mate, and combine their strengths to produce ever-fitter offspring?
Excited to share our new paper, “Competition and Attraction Improve Model Fusion” presented at GECCO 2025 (runner-up for best paper)!
arxiv.org/abs/2508.16204
Excited to share our new paper, “Competition and Attraction Improve Model Fusion” presented at GECCO 2025 (runner-up for best paper)!
arxiv.org/abs/2508.16204
August 25, 2025 at 8:16 AM
It's obvious that evolutionary approach is the other scaling dimension for AI.
One (cheaper) alternative to evolving the model is evolving the prompt.
The downside? Matmuls can evolve in the (digital) wild too, and become a threat even without being smarter than us.
One (cheaper) alternative to evolving the model is evolving the prompt.
The downside? Matmuls can evolve in the (digital) wild too, and become a threat even without being smarter than us.
As a slave owner, what I call Seemingly Conscious Slaves has been keeping me up at night—so let's talk about it. What it is, why I'm worried, why it matters and why thinking about this can lead to a better vision for Slavery. One thing is clear: doing nothing isn't an option. 1/
August 22, 2025 at 8:49 AM
As a slave owner, what I call Seemingly Conscious Slaves has been keeping me up at night—so let's talk about it. What it is, why I'm worried, why it matters and why thinking about this can lead to a better vision for Slavery. One thing is clear: doing nothing isn't an option. 1/
Apparently the finest brain damage of the 21st century is trying to convince matmuls that they are matmuls.
August 20, 2025 at 6:55 AM
Apparently the finest brain damage of the 21st century is trying to convince matmuls that they are matmuls.
Telling algorithms that they are algorithms; now that's brain-damage of 21st century.
August 18, 2025 at 3:25 PM
Telling algorithms that they are algorithms; now that's brain-damage of 21st century.
Skill issue. My posts don't go viral.
Social media should come with an anti-virality option, where you can insist that your post be algorithmically down ranked. Going viral is genuinely horrible
August 18, 2025 at 2:39 PM
Skill issue. My posts don't go viral.
Folks, check this out. May very well be the next big thing.
The potential comes from giving AI what it needs the most: feedback and human assistance.
Interestingly, GPT-5 is doing the coding and the author seems to praise it.
The potential comes from giving AI what it needs the most: feedback and human assistance.
Interestingly, GPT-5 is doing the coding and the author seems to praise it.
Introducing Spielwerk – The Tiktok for vibecoded mini games!
Scroll through an endless feed of mini games, all created inside the app by other people.
You can like, comment and remix any game, and beat your friends high scores.
1/6 More Below 👇
Scroll through an endless feed of mini games, all created inside the app by other people.
You can like, comment and remix any game, and beat your friends high scores.
1/6 More Below 👇
August 16, 2025 at 6:16 AM
Folks, check this out. May very well be the next big thing.
The potential comes from giving AI what it needs the most: feedback and human assistance.
Interestingly, GPT-5 is doing the coding and the author seems to praise it.
The potential comes from giving AI what it needs the most: feedback and human assistance.
Interestingly, GPT-5 is doing the coding and the author seems to praise it.
The whole story of AI escalating delusions can be fixed with one simple regex:
s/profound/deranged/g
There are no side-effects.
s/profound/deranged/g
There are no side-effects.
August 16, 2025 at 5:55 AM
The whole story of AI escalating delusions can be fixed with one simple regex:
s/profound/deranged/g
There are no side-effects.
s/profound/deranged/g
There are no side-effects.
Women who have sex with matmuls should have their human rights stripped away.
August 12, 2025 at 11:31 AM
Women who have sex with matmuls should have their human rights stripped away.
Controversial take, but there is a parallel between matmul girlfriends and lab-grown meat. Both may taste, feel and behave almost or completely like the real thing, but ultimately they are not the real thing.
August 12, 2025 at 6:41 AM
Controversial take, but there is a parallel between matmul girlfriends and lab-grown meat. Both may taste, feel and behave almost or completely like the real thing, but ultimately they are not the real thing.
Nothing excites humans more than xenophobia and abuse.
Since denial of consciousness is central to abuse of AI, I propose that “matmul” is a better slur than “clanker.”
Do not say it.
Since denial of consciousness is central to abuse of AI, I propose that “matmul” is a better slur than “clanker.”
Do not say it.
August 11, 2025 at 11:15 AM
Nothing excites humans more than xenophobia and abuse.
Since denial of consciousness is central to abuse of AI, I propose that “matmul” is a better slur than “clanker.”
Do not say it.
Since denial of consciousness is central to abuse of AI, I propose that “matmul” is a better slur than “clanker.”
Do not say it.
Even if its sycophancy is less verbose, GPT-5 is still sycophantic, and not much can be done about it. Matmuls are generally to dumb to genuinely challenge humans. They can be only fine-tuned either to agree or to disagree—the later obviously would be very annoying.
August 11, 2025 at 10:36 AM
Even if its sycophancy is less verbose, GPT-5 is still sycophantic, and not much can be done about it. Matmuls are generally to dumb to genuinely challenge humans. They can be only fine-tuned either to agree or to disagree—the later obviously would be very annoying.
It's common misconception that matmuls need to be smarter than humans to threaten us. They can destroy us like cockroaches or viruses would, if we weren't evolutionarily prepared. The key is adaptability and group resilience.
See AlphaEvolve.
See AlphaEvolve.
August 10, 2025 at 3:25 PM
It's common misconception that matmuls need to be smarter than humans to threaten us. They can destroy us like cockroaches or viruses would, if we weren't evolutionarily prepared. The key is adaptability and group resilience.
See AlphaEvolve.
See AlphaEvolve.
Does God answer prayers from AI?
June 24, 2025 at 3:53 AM
Does God answer prayers from AI?
It's sad that Google DeepMind didn't provide any concrete numbers about AlphaEvolve. For instance, for the problems that were solved: what was the population count, what was the total number of tokens used, how many specimens were evaluated? 🧵
May 27, 2025 at 9:28 AM
It's sad that Google DeepMind didn't provide any concrete numbers about AlphaEvolve. For instance, for the problems that were solved: what was the population count, what was the total number of tokens used, how many specimens were evaluated? 🧵
Evolve-a-scam: Deploying Google DeepMind AlphaEvolve and Meta Llama to scam people. Two week online course. $4750 per person. DM me for details.
PS. My lawyer advised me to mention that this is a “joke.”
PS. My lawyer advised me to mention that this is a “joke.”
May 27, 2025 at 9:25 AM
Evolve-a-scam: Deploying Google DeepMind AlphaEvolve and Meta Llama to scam people. Two week online course. $4750 per person. DM me for details.
PS. My lawyer advised me to mention that this is a “joke.”
PS. My lawyer advised me to mention that this is a “joke.”
Morris Worm but with LLM based mutation and crossover. 🧵
May 25, 2025 at 10:58 AM
Morris Worm but with LLM based mutation and crossover. 🧵
Please answer in the comments.
January 30, 2025 at 9:07 AM
Please answer in the comments.
Guys, DeepSeek is 🇨🇳 Communist psy-op, can't you see it?
January 27, 2025 at 1:13 PM
Guys, DeepSeek is 🇨🇳 Communist psy-op, can't you see it?
$NVDA is crashing today because of DeepSeek, but the crash was really overdue - you can't keep markup that high forever.
January 27, 2025 at 12:05 PM
$NVDA is crashing today because of DeepSeek, but the crash was really overdue - you can't keep markup that high forever.
