Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
LLM とプロンプトエンジニアリング/チューターをビルドする / LLM, Prompt En...
Search
Kenji Saito
PRO
April 11, 2025
Technology
1
400
LLM とプロンプトエンジニアリング/チューターをビルドする / LLM, Prompt Engineering and Building Tutors
早稲田大学大学院経営管理研究科「プロンプトエンジニアリング ─ 生成AI の応用」2025 春の第1-2回で使用したスライドです。
Kenji Saito
PRO
April 11, 2025
Tweet
Share
More Decks by Kenji Saito
See All by Kenji Saito
ロボットを雰囲気(ヴァイブ)でプログラミングするこどもたち / Children Vibe-Programming Robots
ks91
PRO
0
6
アカデミーキャンプ 2025 SuuuuuuMMeR「燃えろ!!ロボコン」 / Academy Camp 2025 SuuuuuuMMeR "Burn the Spirit, Robocon!!" DAY 3
ks91
PRO
0
13
アカデミーキャンプ 2025 SuuuuuuMMeR「燃えろ!!ロボコン」 / Academy Camp 2025 SuuuuuuMMeR "Burn the Spirit, Robocon!!" DAY 2
ks91
PRO
0
22
アカデミーキャンプ 2025 SuuuuuuMMeR「燃えろ!!ロボコン」 / Academy Camp 2025 SuuuuuuMMeR "Burn the Spirit, Robocon!!" DAY 1
ks91
PRO
0
130
未来へのフォワードキャスト / Forward Cast to the Future
ks91
PRO
0
76
発表と総括 / Presentations and Summary
ks91
PRO
0
57
サイバーフィジカル社会、金融の未来とアイデアソン / Cyber Physical Society, Future of Finance, and Ideathon
ks91
PRO
0
75
マニフェスト: 人類の知のフロンティアに向けた拡張的足場へ / Manifesto: Toward Expansive Scaffolding for Humanity's Knowledge Frontier
ks91
PRO
0
36
続・スマートコントラクトと分散ファイナンス / Smart Contracts and Decentralized Finance, Continued
ks91
PRO
0
75
Other Decks in Technology
See All in Technology
AIに目を奪われすぎて、周りの困っている人間が見えなくなっていませんか?
cap120
1
540
Bet "Bet AI" - Accelerating Our AI Journey #BetAIDay
layerx
PRO
4
1.7k
Segment Anything Modelの最新動向:SAM2とその発展系
tenten0727
0
660
プロダクトエンジニアリングで開発の楽しさを拡張する話
barometrica
0
140
React Server ComponentsでAPI不要の開発体験
polidog
PRO
0
170
AWS DDoS攻撃防御の最前線
ryutakondo
1
150
ホリスティックテスティングの右側も大切にする 〜2つの[はか]る〜 / Holistic Testing: Right Side Matters
nihonbuson
PRO
0
670
AI関数が早くなったので試してみよう
kumakura
0
250
2025-07-31: GitHub Copilot Agent mode at Vibe Coding Cafe (15min)
chomado
2
400
Findy Freelance 利用シーン別AI活用例
ness
0
420
AI時代の経営、Bet AI Vision #BetAIDay
layerx
PRO
1
1.9k
Telemetry APIから学ぶGoogle Cloud ObservabilityとOpenTelemetryの現在 / getting-started-telemetry-api-with-google-cloud
k6s4i53rx
0
140
Featured
See All Featured
The Art of Programming - Codeland 2020
erikaheidi
54
13k
How GitHub (no longer) Works
holman
314
140k
What’s in a name? Adding method to the madness
productmarketing
PRO
23
3.6k
Thoughts on Productivity
jonyablonski
69
4.8k
The Pragmatic Product Professional
lauravandoore
36
6.8k
Connecting the Dots Between Site Speed, User Experience & Your Business [WebExpo 2025]
tammyeverts
8
440
How To Stay Up To Date on Web Technology
chriscoyier
790
250k
Designing for humans not robots
tammielis
253
25k
Keith and Marios Guide to Fast Websites
keithpitt
411
22k
The Illustrated Children's Guide to Kubernetes
chrisshort
48
50k
Improving Core Web Vitals using Speculation Rules API
sergeychernyshev
18
1.1k
Balancing Empowerment & Direction
lara
1
530
Transcript
Generated by Stable Diffusion 3.5 — AI 2025 1-2 LLM
(WBS) 2025 1-2 LLM — 2025-04-14 – p.1/44
https://speakerdeck.com/ks91/collections/prompt-engineering-2025-spring 2025 1-2 LLM — 2025-04-14 – p.2/44
( ) ( ) ( ) 1993 ( ) 2006
( ) SFC 24 P2P (Peer-to-Peer) 2011 ( ) 2018 2019 VR 2022.3 2023 AI VR&RPG 2023.5 “Don’t Be So Serious” 2023 2024 AI( ) 2024 “ALOHA FROM HAWAII” 2024 AI( ) 2025 (Live and Let Think) 2025 Secret Vision Institute ( ) → ( ) 2025 1-2 LLM — 2025-04-14 – p.3/44
Dropbox Dropbox ( ) ( ) ( ) 2025 1-2
LLM — 2025-04-14 – p.4/44
LLM ( ) GPTs 2025 1-2 LLM — 2025-04-14 –
p.5/44
(LLM : Large Language Model) AI (AGI : Artificial General
Intelligence) ( ) ( . . . ^^;) 2025 1-2 LLM — 2025-04-14 – p.6/44
( ) LLM LLM LLM ⇒ 2025 1-2 LLM —
2025-04-14 – p.7/44
( ) MBA MBA · MBA ↑ MBA 2025 1-2
LLM — 2025-04-14 – p.8/44
( ) 1 4 14 LLM ( ) • 2
4 14 ( ) • 3 4 21 I ( ) 4 4 21 ( ) 5 4 28 6 4 28 7 5 12 8 5 12 9 5 19 II 10 5 19 (1) 11 5 26 (2) 12 5 26 (3) 13 6 2 & ( ) 14 6 2 & ( ) W-IOI 2025 1-2 LLM — 2025-04-14 – p.9/44
( 20 ) 1 • 2 • 3 • 4
5 6 RPG 7 “September 12th” 8 9 10 ∼ 11 Linux (Windows )(Mac ) 12 Open Interpreter ∼ 13 01 ∼ 14 AGI (Artificial General Intelligence) 7 (4/28 ) / (2 ) OK / 2025 1-2 LLM — 2025-04-14 – p.10/44
. . . . . . ( ) ( 20
×14 ) 2025 1-2 LLM — 2025-04-14 – p.11/44
(2 )(160 ) (10∼20 ) and/or I (60∼80 ) II
(60∼80 ) 2025 1-2 LLM — 2025-04-14 – p.12/44
Moodle ( Q&A ) ( ) Discord ( ) ←
( ) 2025 1-2 LLM — 2025-04-14 – p.13/44
= A ( ) B ( ) A A (
) 2025 1-2 LLM — 2025-04-14 – p.14/44
: OpenAI Playground / Dashboard AI : Discord GPT-Discord Bot
System AI GPT ChatGPT Plus Claude ( LLM) 2025 1-2 LLM — 2025-04-14 – p.15/44
Discord web 60 ⇒ 2 ( ) ( ) 2025
1-2 LLM — 2025-04-14 – p.16/44
LLM ( ) . . . ( ) 2025 1-2
LLM — 2025-04-14 – p.17/44
B. , , 1968 = ⇒ ( ) ( )
AI ( ) 2025 1-2 LLM — 2025-04-14 – p.18/44
: ⇒ = 30 . . . ( ) 30
⇒ + LLM (2023 3 ) + LLM ( ) 2025 1-2 LLM — 2025-04-14 – p.19/44
m P(w1 , . . . , w m )
(Wikipedia) 1 (Wikipedia) : ( ) ← (Generative Pre-training) : ( ) ( ) https://writings.stephenwolfram.com/2023/02/what-is-chatgpt-doing-and-why-does-it-work/ 2025 1-2 LLM — 2025-04-14 – p.20/44
ChatGPT GPT ChatGPT GPT OpenAI (GPT-3.5, GPT-4(.5), o1, o3, etc.)
GPT Generative Pre-trained Transformer ( ) GPT GPT ( ) GPT-3.5, GPT-4 2025 1-2 LLM — 2025-04-14 – p.21/44
GPT 2025 1-2 LLM — 2025-04-14 – p.22/44
ChatGPT ChatGPT ← LLM LLM → 200 ( ) 2025
1-2 LLM — 2025-04-14 – p.23/44
ChatGPT ( 4.5) ChatGPT – Deep Research OpenAI Playground (gpt-4o)
Google ( ) Perplexity Grok Claude 2025 1-2 LLM — 2025-04-14 – p.24/44
GPT 2025 1-2 LLM — 2025-04-14 – p.25/44
( ) GPT 3 1. (GPT ) 2. ( )
3. ( ) 3 (↑ 3. ) 3. (GPT ) 2025 1-2 LLM — 2025-04-14 – p.26/44
GPTs ( ) ( ) 2025 1-2 LLM — 2025-04-14
– p.27/44
GPTs ʂ (GPT) ChatGPT 2025 1-2 LLM — 2025-04-14 –
p.28/44
GPT Ζ͏ͥ + 2025 1-2 LLM — 2025-04-14 – p.29/44
ͬͪ͜ͰϏϧμʔͱର͠ͳ͕Β ৽͍͠ (15 ΛΕ·͢ ͬͪ͜Ͱ৽͍͠ (15 Λࢼͤ·͢ ͲΜͳϓϩϯϓτʹͳ͍ͬͯΔ͔ΛݟͨΓ ࡉ͔͘मਖ਼ͨ͠Γઃఆͨ͠ΓͰ͖·͢ GPT
2025 1-2 LLM — 2025-04-14 – p.30/44
( ) GPT 3 1. (GPT ) 2. ( )
3. ( ) 3 (↑ 3. ) 3. (GPT ) 2025 1-2 LLM — 2025-04-14 – p.31/44
PAT (Paper Authoring Tutor) : GPTs : https://chatgpt.com/g/g-6lW16zlHo-hatuto-pat-lun-wen-asisutanto GitHub :
https://github.com/ks91/pat GPTs PAT PAT 2025 1-2 LLM — 2025-04-14 – p.32/44
ίϯηϓτΛ࿅Δɾ "* ͱ ૬ஊ͢Δ ૬ஊ͢Δɾߟ͑Δ ର͢Δ ͪΐͬͱ ҧ͏Μͩͳʔ Α͠ࢼͯ͠ΈΑ͏ ࢥͬͯͨͷͱ
ҧ͏ Α͠ ࡞ͬͯ ΈΑ͏ ߟ͑͢ ࢦࣔ ࢿྉΛ ೖྗ͢Δ ΞγελϯτΛ࡞Γ ߋ৽͢Δ ΞγελϯτΛ ࢼ͢ 2025 1-2 LLM — 2025-04-14 – p.33/44
( ) : https://diamond.jp/articles/-/360966 ( ) : https://diamond.jp/articles/-/360967 AGI 2025
1-2 LLM — 2025-04-14 – p.34/44
AI ࣗಈԽ γϛϡϨʔγϣϯ ਓؒͷ֦ு ϓϩϯϓτ ΤϯδχΞϦϯά ੜ "* ˞͔͋ͨ࿑ಇऀΛ૿͢ ˞ਓؒʹ࡞༻্ͤ͠͞Δ
˞ϞσϧͰ༧ଌ͠౷ޚ͢Δ େͷ Φʔτϝʔγϣϯ খɾےͷ Φʔτϝʔγϣϯ ਓମͷ࡞༻ σδλϧϨΠόʔ σδλϧπΠϯ αΠόʔϑΟδΧϧ ػցֶश ༧ҩྍ γϦΞεήʔϜ Ϗοάσʔλ ςϨΤάδελϯε ϦϞʔτϫʔΫ Ծɾ֦ுݱ࣮ Y3 ࠶ੜҩྍ #.* ৽ༀ ԕִҩྍ ϩϘςΟΫε ݴ༿Ͱ "* ͱର͠ɺΰʔϧͷ ୡʹ͔͏ͨΊͷϓϩϯϓτ ΤϯδχΞϦϯά͕ɺͯ͢ͷ جૅʹҐஔ͚ͮΒΕ͍ͯ͘ ڥٕज़ "* લఏࣾձʹ͓͍ͯɺ ؾީมಈɺਓؒͷपΓͷ ॏେͳʹରͯ͠ ࣗಈԽͱγϛϡϨʔγϣϯ ʹΑཱ͔͍ͬͯͪͬͯ͘ "* લఏࣾձʹ͓͍ͯɺ ΤωϧΪʔ৯ྉΛ͡Ίͱ͢Δ ੜ࢈ྲྀ௨ͷࣗಈԽɺ ৗੜ׆ͷϩϘςΟΫεʹΑΔࢧԉ͕ ࢝·͍͖ͬͯɺ·ͨɺ ϗϫΠτΧϥʔͷ࿑ಇऀ͕ैࣄ͢Δ ࣄͷଟ͘σδλϧϨΠόʔͷํ͕ ͏·͘ΕΔΑ͏ʹͳ͍ͬͯ͘ "* લఏࣾձʹ͓͍ͯɺ ݸਓ͕Τϯύϫʔ͞Εɺ ਓͻͱΓ͕Ͱ͖Δ͜ͱ͕ ֦େ͍ͯ͘͜͠ͱʹՃ͑ɺ ݈߁্͕͠ɺ ߴྸԽΛՃͤͭͭ͞ ࣗಈԽʹΑΓ શһ͕࿑ಇਓޱ͔Β ୀग़Λ࢝Ί͍ͯ͘ "* લఏࣾձʹ͓͍ͯɺ ͔͋ͨશਓྨͷ ༧ଌɾੳೳྗ্͕͢Δ 2025 1-2 LLM — 2025-04-14 – p.35/44
AGI AGI : Artificial General Intelligence ( ) ( )
( AGI ) [ AGI] “Star Trek: The Next Generation” ( ) (1) ( ) ( ) Jean-Luc Picard “Computer, tea. Earl Grey. Hot.” ( ) (2) ( ) ( ) ( ) (1) 2025 1-2 LLM — 2025-04-14 – p.36/44
AGI 5 (OpenAI ) 1 (ChatGPT) 2 (o1, o3) 3
( ) (> deep research) (A ) 4 ( ) (X ) 5 (AGI ) 3 5 A X 2025 1-2 LLM — 2025-04-14 – p.37/44
A X — A M. A /X MIT 2013 LEGO
X A A X A X X X (2013 ) ⇒ 10 A Open Interpreter ChatGPT Deep Research A . . . 2025 1-2 LLM — 2025-04-14 – p.38/44
(Amodei, 2024) 5 (AGI ) 1. AI 10 21 5∼10
2 2. AI 3. AI AI 4. 5. (AI ) 2025 1-2 LLM — 2025-04-14 – p.39/44
( ) 6 10 ( ) — ( ; )
Saito Lab at WBS 1 ( ) (1) ( ) (2) OpenAI (ChatGPT OK) (3) ( ) Moodle OpenAI 2025 1-2 LLM — 2025-04-14 – p.40/44
( ) Code interpreter File Search ( OK) ( )
2025 1-2 LLM — 2025-04-14 – p.41/44
2025 1-2 LLM — 2025-04-14 – p.42/44
1. (1) AI (2) 2025 4 17 ( ) 23:59
JST ( ) Waseda Moodle 2025 1-2 LLM — 2025-04-14 – p.43/44
2025 1-2 LLM — 2025-04-14 – p.44/44