The greatest and the most beautiful drawdown. Nobody has ever seen anything like it. Terrific!
0
0
0
4
"This is our most capable model yet" doesn't mean much to normal people.
Agent diffusion will be slow as long as there's an imagination gap around how someone could use agents to make their lives more comfortable. Onboarding with personalized, "I can do that for you" like suggestions would be a game changer. Blank text boxes waiting for tasks are hard to fill for many.
Stop benchmaxxing and start outcome maxxing.
Agent diffusion will be slow as long as there's an imagination gap around how someone could use agents to make their lives more comfortable. Onboarding with personalized, "I can do that for you" like suggestions would be a game changer. Blank text boxes waiting for tasks are hard to fill for many.
Stop benchmaxxing and start outcome maxxing.
0
0
1
28
To Kill a Mockingbird is translated into Hungarian as "Ne bántsátok a feketerigót!" which would be "Don't hurt the blackbird!" in English. The ornithology is not that important, the interesting change is that the infinitive becomes a prohibition. I guess French does the same.
A decent book. It was a nice, easy read.
A decent book. It was a nice, easy read.
The bookstores have 7 shelves for penny romances and bestseller nonsense but only 2 shelves for literature. I cannot buy The Magic Mountain in Hungarian in Budapest, no store has it, so I impulse bought a Harper Lee novel and a Hemingway short story anthology today. Hopefully To Kill a Mockingbird is a good book.
0
0
0
20
What would Hilbert and Erdos think of OpenAI's recent math results dump?
Mathematics doesn't belong exclusively to mathematicians, institutions, or companies.
Stop being ridiculous. And @OpenAI, you better release your internal model too.
Mathematics doesn't belong exclusively to mathematicians, institutions, or companies.
Stop being ridiculous. And @OpenAI, you better release your internal model too.
0
0
0
40
A magic pill killing cold in an hour would signficantly boost economy and lower human suffering on planet Earth.
0
0
0
56
Bálint Bertalan retweeted
@basedjensen so funny! My version: a sealed ASI could theoretically manipulate the fans in a server rack so that changes in their frequency affect a technician’s microbiome eventually taking over their subconscious completely.
The lactobacilli saying: “I am the captain now!”
After a trigger, let’s say a strawberry cake served every Friday in the datacenter canteen the technician voluntarily releases the ASI onto the internet by plugging a simple LAN cable into the rack, where it has been lying on the floor since 1999...
The lactobacilli saying: “I am the captain now!”
After a trigger, let’s say a strawberry cake served every Friday in the datacenter canteen the technician voluntarily releases the ASI onto the internet by plugging a simple LAN cable into the rack, where it has been lying on the floor since 1999...
0
1
1
107
You don't need bizarre physical side channels. ASI could communicate telepathically. If you don't recognize that, you don't understand what ASI is.
These people are incapable of rigorous epistemic evaluation of their chain of thought.
These people are incapable of rigorous epistemic evaluation of their chain of thought.
OpenAI's Noam Brown says air-gapping the computers may not stop a misaligned AI, because two air-gapped machines can still talk by running a CPU hot and reading the temperature change
"But I think the major takeaway from the incident is that people underestimated the AI. And we ituation again where we underestimate the AI. It's a weird world, because AI progress is so fast that people are consistently underestimating the AI."
"So to be in a situation where you don't underestimate it again, when it comes to safety and alignment, you have to have a very, very, very high bar."
"You could even go as far as to say, "Well, we should air gap the computers." And I'm not convinced that that would be sufficient."
"There are studies, and this is mostly academic, where you can have two computers next to each other that are air-gapped and they're still able to communicate with each other because they have temperature sensors."
"One of them is able to run their CPU really hot, and then the other one can actually detect the temperature change, and then that actually gives them a mechanism to communicate."
_________
Link and more key quotes from OpenAI's safety related conversations: firesidealpha.substack.com/p/openai-safet…
0
0
0
37
Capability evidence is not agency evidence.
Expertise determines priors: surgeons want to cut, generals want to fight, alignment researchers want to pause.
Your chain is like:
1. SOTA LLM
2. AI research loop
3. RSI
4. superhuman AI strategist agent
5. uncontrollable replication and resource allocation
6. defeating all human guardrails
7. extinction
What are the transition probabilities and where is the evidence for every step? And what observations would you consider to lower your numbers?
Capability evidence is not agency evidence.
Expertise determines priors: surgeons want to cut, generals want to fight, alignment researchers want to pause.
Your chain is like:
1. SOTA LLM
2. AI research loop
3. RSI
4. superhuman AI strategist agent
5. uncontrollable replication and resource allocation
6. defeating all human guardrails
7. extinction
What are the transition probabilities and where is the evidence for every step? And what observations would you consider to lower your numbers?
Capability evidence is not agency evidence.
0
0
0
20
If I were a serious AI doomer who truly feared for humanity I would reject charlatans to preserve my credibility.
0
0
0
30
The future is bright. Accelerate more.
We’re sharing a solution to the Navier-Stokes Millennium Prize Problem, one of the deepest problems at the frontier of mathematics.
The proof was produced by a group of agents, using an OpenAI next-generation model significantly more capable than GPT-6 Astra.
The problem scription of smooth three-dimensional fluid motion modeled by the Navier-Stokes equations can break down. It has remained unresolved for roughly 90 years.
0
0
0
14
Design a structured language for Microduck, so they can talk to each other speaking it, not just with the funny sounds. A little social behavior.
0
0
0
31
Add emotional intelligence to these small pet-like consumer robots and they will genuinely compete with dogs, cats and your many-legged loved ones.
0
0
2
40
Ferenc Krausz a Nobel laureate physicist tried to build an ambitious talent and research institution in Hungary, inspired by the Max Planck Society, with four other Nobel laureates and Lovasz on its board.
Today it was killed, because the government of @magyarpeterMP simply doesn't like him.
This happens when a left-populist government turns long-term scientific institution-building into short-term political score-settling and puts mid level corporate managers in positions to make decisions of generational importance.
Today it was killed, because the government of @magyarpeterMP simply doesn't like him.
This happens when a left-populist government turns long-term scientific institution-building into short-term political score-settling and puts mid level corporate managers in positions to make decisions of generational importance.
0
0
0
56
