mindgitrwx @mindgitrwx

I use this twitter account as an index table Joined December 2017

Tweets

1K
Followers

142
Following

5K
Likes

17K

Derek Thompson @DKThomp

2 months ago

Yes. Writing is not a second thing that happens after thinking. The act of writing is an act of thinking. Writing *is* thinking. Students, academics, and anyone else who outsources their writing to LLMs will find their screens full of words and their minds emptied of thought.

541 7K 30K 2.4M 13K

Download Image

Simon Willison @simonw

2 months ago

Wrote some notes on Toad, Will's new not-yet-open-source-but-soon terminal coding agent built on Textual He's charging companies $5,000 for early access to the preview before it goes open source, I'd love to see that model work here! simonwillison.net/2025/Jul/23/an…

Will McGugan @willmcgugan

2 months ago

58 76 678 92K 427

Download Image

1 19 171 21K 60

GeekNews @GeekNewsHada

2 months ago

왜 사실만으로는 생각이 바뀌지 않는가 — "구조가 답이다" - 갈릴레오의 지동설 논쟁 사례처럼, 사회적 권력과 신념 체계는 단순한 '사실'만으로 바뀌지 않음 - 교회·권력·이념은 성경, 우주관, 예술, 사회 규범 등 서사와 구조적 연결을 통해 현실을 설명하… news.hada.io/topic?id=22125

0 11 25 2K 16

jason liu @jxnlco

2 months ago

genuinely the coolest blog ever: ampcode.com/how-to-build-a… sending this to an agent to make your first agent is nuts

20 280 3K 192K 5K

alphaXiv @askalphaxiv

2 months ago

"How Many Instructions Can LLMs Follow at Once?" In this paper they found that leading LLMs can satisfy only about 68% of 500 concurrent instructions, showing a bias toward earlier instructions.

14 150 838 93K 596

Download Image

말러팔삼 @mahler83

2 months ago

뇌는 주어진 정보로 다음 순간을 끊임없이 예측하고 있고, 그것이 틀렸을 때 강력한 학습을 일으키는 도파민 신호가 나오고, 다음에는 맞출 수 있게 된다. "오류 기반 학습"(error-based learning)이 효율적인 이유. pretest, 생산적 실패, 오답분석. LLM 쓸 때도 먼저 시도해야 하는 이유! #학습과학

2 80 197 175K 110

말러팔삼 @mahler83

2 months ago

그거 구글 치면/챗지피티한테 물어보면 다 나오는데, 왜 굳이 외워야 하냐? 지식이 많아질수록 스키마 형태로 통합해서 덩어리로 인식해야 작업기억에서 하나의 단위로 다룰 수가 있음. 하나하나 찾아보고 확인한 낱개 단위로는 작업기억 용량이 딸려 크고 복잡한 사고를 할 수가 없음

말러팔삼 @mahler83

2 months ago

1 132 172 63K 67

2 478 658 167K 301

Simon Willison @simonw

2 months ago

I figured out how to add the official Playwright browser automation MCP to Claude Code. Run this before you start "claude": claude mcp add playwright npx '@playwright/mcp@latest' Now Claude Code can use a Chrome browser directly! Here's my TIL: til.simonwillison.net/claude-code/pl…

43 91 1K 125K 2K

Neel Nanda @NeelNanda5

2 months ago

I'm very excited about our vision for "mech interp" of CoT: Study reasoning steps and their connections - analogous to activations Don't just read it: study attn, causally intervene, and, crucially, resampling - study the distn over CoTs, not just this one There's lots to do!

Paul Bogdan @paulcbogdan

2 months ago

18 152 772 119K 840

Download Video

4 62 556 50K 440

Download Image

Paul Bogdan @paulcbogdan

2 months ago

New paper: What happens when an LLM reasons? We created methods to interpret reasoning steps & their connections: resampling CoT, attention analysis, & suppressing attention We discover thought anchors: key steps shaping everything else. Check our tool & unpack CoT yourself 🧵

18 152 772 119K 840

Download Video

Andrej Karpathy @karpathy

2 months ago

+1 for "context engineering" over "prompt engineering". People associate prompts with short task descriptions you'd give an LLM in your day-to-day use. When in every industrial-strength LLM app, context engineering is the delicate art and science of filling the context window…

tobi lutke @tobi

3 months ago

354 894 9K 2.0M 2K

535 2K 14K 2.3M 9K

alphaXiv @askalphaxiv

3 months ago

Introducing your arXiv Research Agent A personal research assistant with access to arXiv + bioRxiv + medRxiv + Semantic Scholar. Upload drafts, conduct literature reviews, get insights across millions of papers MCP support coming soon 🚀

25 303 2K 120K 1K

Download Video

jason liu @jxnlco

4 months ago

i'm watching rag systems evolve from "chat with your pdfs" into something far more valuable. the problem? most rag implementations are primitive text chunking exercises that miss the real economic opportunity. here's what actually matters: extract structured data, not just…

37 64 1K 168K 2K

Simon Willison @simonw

4 months ago

I put together an annotated version of the new Claude 4 system prompt, covering both the prompt Anthropic published and the missing, leaked sections (thanks, @elder_plinius) that describe its various tools It's basically the secret missing manual for Claude 4, it's fascinating!

39 369 4K 685K 6K

Download Image

Kairos💜🩷 @AshesofTime

4 months ago

“짧은 글이나 단어를 일주일에 다섯 번씩 소리 내어 릵는 간단한 프로그램이었다. 그 외에 다른 변수는 없었다. 그런데 놀랍게도, 이 훈련만으로도 치매 환자들의 인지 기능이 향상됐다.”

1 444 781 45K 525

Download Image

말러팔삼 @mahler83

4 months ago

재미있는 #논문 이네. 사람들이나 AI알고리즘이나 둘다 블랙박스인데 사람의 결정은 자신이 더 잘 이해한다고 착각함. 이 연구에서 사용한 사례는 판사의 판결, 면접관의 채용 결정, 영상의학의사의 판독 3가지. "그럼 설명해보세요"라고 하면 본인이 이해하고 있다고 믿는 점수가 뚝 떨어짐 ㅋㅋ

2 132 226 17K 124

Download Image

Stephen Wolfram @stephen_wolfram

6 months ago

From the point of view of science I've never been too excited about the Game of Life. But now I realize that what's really exciting about it is what it tells us about what we can call metaengineering. That half century of impressive Life hacking gives us a uniquely clean example…

41 240 2K 94K 759

Download Image

vittorio @IterIntellectus

6 months ago

cognitive decline at 30 is a lie your skills don't peak until your mid-40s and even then, losing them is entirely your choice. 1/

270 1K 15K 1.0M 6K

Download Image

Nauseam @ChadNauseam

7 months ago

"A calculator app? Anyone could make that." Not true. A calculator should show you the result of the mathematical expression you entered. That's much, much harder than it sounds. What I'm about to tell you is the greatest calculator app development story ever told.