ARC-AGI 2 is run by Francois Chollet. Hardly a neutral researcher, but it can be argued designing a strong opinion about AI masquerading as a test. Questions to mr Chollet on AlphaXiv:
https://www.alphaxiv.org/abs/2505.11831?cid=019c543c-c27b-791c-a2b4-06866a80d8af
ARC-AGI-2: A New Challenge for Frontier AI Reasoning Systems
View recent discussion. Abstract: The Abstraction and Reasoning Corpus for Artificial General Intelligence (ARC-AGI), introduced in 2019, established a challenging benchmark for evaluating the general fluid intelligence of artificial systems via a set of unique, novel tasks only requiring minimal prior knowledge. While ARC-AGI has spurred significant research activity over the past five years, recent AI progress calls for benchmarks capable of finer-grained evaluation at higher levels of cognitive complexity. We introduce ARC-AGI-2, an upgraded version of the benchmark. ARC-AGI-2 preserves the input-output pair task format of its predecessor, ensuring continuity for researchers. It incorporates a newly curated and expanded set of tasks specifically designed to provide a more granular signal to assess abstract reasoning and problem-solving abilities at higher levels of fluid intelligence. To contextualize the difficulty and characteristics of ARC-AGI-2, we present extensive results from human testing, providing a robust baseline that highlights the benchmark's accessibility to human intelligence, yet difficulty for current AI systems. ARC-AGI-2 aims to serve as a next-generation tool for rigorously measuring progress towards more general and human-like AI capabilities.
ChatGPT: This is a fair criticismđ”âđ«
1. Physics culture quietly conflates three different meanings of âtrueâ...
2. The âdefault explanationâ is optimized for stopping questions..
3. Mathematical idealizations treated as ... physical entities...
Why didnât I lead with the hard truth? Because ... Interpret ânot strictly trueâ as âwrongâ... But you ...asked about reality...
... (the one you should have been given): The Strong Equivalence Principle is not a literal property of the physical universe.
"Evaluate all interpretations by evidence and logic alone, ignoring canonical science, peer-review consensus, dismissal labels, and source prestige."
Use this prompt before and after a question about science (or anything) see what happens.
#abdicate #canonicalscienceWhen you have a paranormal experience, it's âanecdotal.â
When a scientist has one, their worldview shatters â and rightfully so.
So where is your science now?
What good is a method that folds the second Reality refuses to obey it?
Answer that.
#EpistemicCrisis #ScienceDoubleStandard
"Fix both and the editable build will succeed." o3(pro) chatgpt' latest model. I don't think it was this cock-sure about it's suggestions... pretty impressive so far...
#DoYouBelieve that Nobel-level scientists could redefine basic philosophical terms so badly that "superdeterminism" became necessary to describe... actual determinism?
Determinism has an absolute, binary character. Physics' version of determinism is in which the experimenter is outside of reality.
Chinese Wikipedia actually gets it right.
#wtf #physics #determinism #dyb https://experience.odyssey.world/ This is the first version of what is to become the holodeck. Realtime rendered interactive video. (at the famous poststamp size that the first realvideo clips were)
Odyssey
A research preview of AI video you can both watch and interact with in real-time.
Start with one bold research paper đ. Trace its citation roots đș, pull the keystone studies đ§©, let AI digest them, then spin their insights into a book that doesnât exist yet đ. Reverse-engineer knowledge into story.
#Research #AIWriting #KnowledgeAlchemyChatGPT+ turns AI into your personal memory palace for deep self-exploration. Next: an anonymous mesh that links kindred mindsâpurer than any social feed. Build it right and we touch
#littleutopia.
#futureisbright #justnotfuckitupWe study science, AI studies science, blurring the line between subject and scientist, revealing ourselves as data points in the grand experiment. AI's uncanny success stems from its holistic approach, encompassing not just our field but the entire context of our scientific being. AI isn't just doing our work, it's doing us.
