最近在带入组的本科实习生,发现怎么读论文其实是科研训练里最容易被忽略的一步。
推荐一篇每个科研新人都该读的经典短文:S. Keshav 的 How to Read a Paper。
文章提出了非常实用的“三遍读论文法”:
第一遍,5 到 10 分钟快速扫读:标题、摘要、引言、章节标题、结论和参考文献。
目标是回答 5C:
Category, Context, Correctness, Contributions, Clarity。
也就是判断这篇论文是什么、和谁相关、假设是否合理、贡献是什么、写得清不清楚。
第二遍,认真读论文主线,但先跳过证明细节。重点看图表、实验设置、结果是否清楚、引用了哪些关键工作。
第三遍才进入深度理解:尝试像复现一样重建作者的思路,检查假设、方法、创新点和潜在漏洞。
放在今天看,这个方法和 AI 辅助读论文其实很契合。
第一遍可以让 AI 帮忙快速总结论文的研究问题、核心贡献和主要结论,但自己一定要判断这篇文章是否真的值得继续读。
第二遍可以让 AI 帮忙解释方法、实验设置、图表和不熟悉的概念,但不能只看 AI 总结。关键图表、实验设计和结果数字一定要回到原文核对。
第三遍可以让 AI 扮演 reviewer,帮你追问:这篇文章的假设是否成立?实验是否支持结论?有没有 missing baseline?有没有潜在的数据泄漏、评价偏差或过度 claim?
读论文不是“读完”就行。真正重要的是知道什么时候快速跳过,什么时候认真理解。
尤其在 AI 工具越来越强的情况下,科研新人更需要训练自己的判断力。
AI 可以帮你压缩信息,但不能替你决定一篇论文是否重要、是否可信、是否值得借鉴。
https://t.co/8gUc4HbLwR
Students without access to LLMs are 2 to 8 times more creative than students with access.
That is the finding of a new paper comparing 2,200 college admissions essays written by humans before ChatGPT with essays generated by GPT-4.
The key point is not individual creativity. GPT-4 can write well, sometimes better than individual students. The problem is collective creativity.
Each new human essay added new semantic territory. New ideas. New angles. New experiences. New combinations.
Each new GPT-4 essay added much less.
The authors call this the diversity growth rate: how much novelty each additional text contributes to the collective pool of ideas.
Humans kept expanding the pool. GPT-4 made the pool converge.
Even when the authors pushed GPT-4 to be more creative, changed parameters, or used chain-of-thought prompting, the homogenizing effect remained.
This is the real danger of AI in education.
Not that students will write worse.
That everyone will write the same.
*
Full paper in the first reply
The best Catholic college in America offers only one major, no minors, no technology in the classroom, and they have made absolutely no changes to what they are doing even in the midst of the AI Revolution.
They simply immerse students in beauty - physical beauty, natural beauty, liturgical beauty, intellectual beauty, and relational beauty through deep friendship.
They consistently graduate the most clearheaded, lovely, thoughtful, and normie young people you have ever met. If colleges are going to survive they need to become more like Thomas Aquinas College in California. For those of you who have never seen it before this is what a serious undergraduate education looks like freshman and sophomore year.
Akhirnya saya baru sadar, berada kampus yg ngajarin cara nyari jurnal internasional, ada matkul kepenulisan akademik, ospek diajarin cara bikin sitasi, diajarin etika kepenulisan akademik, dll adalah sebuah privilege di Indonesia 😭😭😭
Making a list of a hundred papers in economics which fill me with wonder, joy, and excitement. Your suggestions are urgently solicited, my memory is not infallible and I'd like to make this really good.
https://t.co/V3rzW1bZgZ
That guy?
Yang 'riset' dia penuh dengan sitasi hasil halusinasi AI???
You want him buat jadi Menteri Ekonomi????
Yang bilang dengan retargeting MBG, PISA score Indonesia bisa naik 3.83% / tahun?????
Mungkin Indonesia pantes bubar kalau gitu.
@arifnovianto_id Kalo gw diajarin dosen malah manfaatin tools ai yang ada
flownya:
1. lempar topik ke ai claude,gpt,dll
2. Minta searching judul dr dianya
3. Hasil judul copas ke researchrabbit ai
4. Tinggal sesuaiin cari yg pas, ada paper yg similar, mengutip/dikutip
Beres daripada paper halu
"berarti kita ga boleh dong pake AI untuk ngerjain tugas/skripsi?".
jawabannya sangat boleh tapi terbatas untuk hal-hal tertentu kyk
-ngejelasin materi yg sulit dengan bahasa sederhana
-ngecek tata bahasa dan kerapian tulisan
-brainstroming dan diskusi
Yang bermasalah adalah langsung minta AI buatin papernya dan copas mentah-mentah
Dear adik2 mahasiswa,
Sangat mudah mendeteksi AI kalau kamu tidak cek ulang makalah kamu padahal gampang sekali melakukan pemeriksaan mandiri. 🧵👇
Buka halaman space qalmsw di huggingface (link cek reply)
upload kode makalah kamu (LaTeX) dan daftar pustakanya (.bib) lalu klik "Run checks"
Nanti ada tabel di bawah yang berisikan "error" seperti di gambar dan saran perbaikannya atau unduh berkas dalam format JSON lalu berikan ke AI agen yang kamu pakai untuk minta perbaikan.
Kalau mau diperiksa lebih jauh pakai AI lagi bisa masukkan API KEY kamu sendiri (BYOK, key ini tidak akan disimpan).
Kalau mau dijalankan sendiri unduh kodenya dari github (link di reply) dan jalankan dengan local AI (llama.cpp)
PS: ini bukan cuma buat mahasiswa tapi Bapak/Ibu dosen juga bisa pakai sebelum submit makalah ke konferensi/jurnal
Short post on the two traditions in the history of economic ideas. The central divide is between classical political economy, reproduction, surplus, accumulation, and the Benthamite/marginalist tradition of utility, exchange, scarcity, and individual choice /1