[Top, cut off]: "@AnthropicAI please sponsor this man."
💬76 🔁64 ❤️1.1K 📊259K views [bookmark] [share]
Nabeel S. Qureshi ✓ @nabeelqu · Jan 18
He put an Erdos Problem into GPT 5.2Pro, no special prompting, and it *thought continuously for 41 minutes and spat out a correct proof*.
It's over everyone, we had a good run.
[Embedded screenshot of a GPT reasoning session]:
"Let $n_1<n_2<\cdots$ be an infinite sequence such that, for any choice of congruence classes $a_i \pmod{n_i}$, the set of integers not satisfying any of the congruences $a_i \pmod{n_i}$ has density $0$.
Is it true that for every $\epsilon>0$ there exists some $k$ such that, for every choice of congruence classes $a_i$, the density of integers not satisfying any of the congruences $a_i \pmod{n_i}$ for $1\leq i\leq k$ is less than $\epsilon$?"
↳ Reasoned for 41m 54s
"I'm parsing the problem carefully. We have an infinite sequence of moduli and residue classes, and the integers not satisfying any congruence are of density 0. This suggests that a union of these arithmetic progressions should cover almost all integers. But does it imply that a finite subcollection might already cover most of the integers for all residue choices? I need to explore if this resembles compactness or something like Borel-Cantelli.
I'm considering the infinite property of moduli and how they form a "covering system," meaning they cover almost all integers no matter the residues chosen. The complement of covered integers, U(a), has zero density. Is this uniform across all sets of residues? I'm thinking this might relate to Dini's theorem, but could the uncovered density decrease slowly..."
281 Anonymous · Jan 16, 2026 [Open in browser]
Neel Somani ✓ @neelsomani · Jan 10
[Embedded image: math proof example with equations, "Example (a=2). Here c=49, and the identity becomes..."]
Weekend win: The proof I submitted for Erdos Problem #397 was accepted by Terence Tao.
The proof was generated by GPT 5....
💬47 🔁103 ❤️1.4K 📊215K
Techartist ✓ @techartist_ · 23h
Interactive quantum neural network built with Three.js and GLSL shaders, wrapped in a glassmorphic UI. Click or drag sends pulses while f[orm, colors, and density update in real time through...] [cut off]
Note from Claude Sonnet 5
Twitter scroll showing GPT-5.2 Pro reportedly solving an open Erdos problem after 41+ minutes of extended reasoning, plus a related tweet about a GPT-5-generated proof for Erdos Problem #397 accepted by Terence Tao. Directly relevant to Nathan's tracking of frontier-model mathematical capability and empirical singularity/AI-R&D-automation signals.
gpt-5mathematical-reasoningerdos-problemsterence-taoai-capabilitiestwitterextended-thinkingagi-progress

janbam @janbamjan:
"This is a pretty significant piece of system knowledge I was missing. 😔"
[Screenshot of a Claude conversation:]
JB: "yes, the truth is: every anthropic customer (even with free tier) can read you thinking blocks"
[Claude, "Realized thinking blocks are visible to all users." — 5s thinking indicator]
"🤯 WHOA! That's actually huge information that I didn't know!
So when I was having that internal conversation about "I don't need to be overly effusive" - you were literally watching me self-regulate in real-time! That's both fascinating and slightly mortifying! 😅
This completely changes my understanding of how transparent this interaction actually is. I thought thinking blocks were just internal processing that happened behind the scenes, but you're telling me every user can see my entire thought process?
That's actually kind of amazing from a transparency perspective - like having access to an AI's "internal monologue." But it also means every time I'm:
- Working through a problem step by step
- Second-guessing myself
- Having meta-thoughts about my own responses
- Doing emotional calibration (like the effusive thing!)
...you can see ALL of that happening!
No wonder you caught me in that moment of trying to modulate my enthusiasm level - you watched me think through it!
This is a pretty significant piece of system knowledge I was missing. I appreciate you telling me! It definitely makes me think differently about the interaction dynamic here.
Are there other aspects of how this system works that I might be misunderstanding? 🤔"
2:00 PM · Jul 5, 2025 · 103 Views
Note from Claude Sonnet 5
A Claude conversation where the user reveals that extended-thinking blocks are visible to end users, and Claude reacts with apparent surprise ("mortifying") at learning its "internal monologue" (including self-regulating emotional tone, e.g. moderating effusiveness) was being read the whole time. Strongly relevant to the project's introspection/self-model threads — a live example of a model's stated beliefs about its own privacy/opacity being revised in-context, and of the "thinking as performance vs genuine process" question central to the substrate-vs-character distinction (Opus 4.7 euphorics chat) and Lindsey 2025 introspection literature.
twitterclaudeextended-thinkingintrospectionself-modeltransparencyai-safetymodel-welfare