Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Sign up for free
Menu
Search
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Features
All features
Private URLs
Password Protection
Custom URLS
Scheduled publishing
Remove Branding
Restrict embedding
Deck Collections
Notes
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Explore
Featured decks
Featured speakers
Programming
Technology
Storyboards
Pricing
Search
Sign in
Sign up for free
生成AIを用いたサービスの紹介
Search
Uryu Shinya
December 05, 2024
Technology
250
1
Share
Embed
Copy iframe code
Copy JS code
Copy link
Start on current slide
生成AIを用いたサービスの紹介
Uryu Shinya
December 05, 2024
More Decks by Uryu Shinya
See All by Uryu Shinya
安全なAI利用のためのLLM(大規模言語モデル)の利用と評価 / japanr2025
s_uryu
0
110
生成AIサービスを用いた研究活動の支援
s_uryu
0
240
R研究集会(2024)のご案内
s_uryu
1
780
生成AIの基礎的事項と社会に与える影響
s_uryu
0
120
Rの機械学習フレームワークの紹介〜tidymodelsを中心に〜 / machine_learning_with_r2024
s_uryu
0
1.6k
地理空間データの機械学習への適用 / machine_learning_for_spatial_data
s_uryu
0
410
mandaRa: R言語ユーザのための新しい知識共有の場 / mandara_tokyor111
s_uryu
2
810
R言語入門 (R-4.3.3 2024年4月版) / introduction to r
s_uryu
7
7.3k
統・再現性・協力: 人為的過誤を防ぎ、未来へ進む策 / Integration, Reproducible, and Collaboration
s_uryu
1
890
Other Decks in Technology
See All in Technology
Refining Active Record
shugo
0
180
Conditional PED-ANOVA: Hyperparameter Importance in Hierarchical & Dynamic Search Spaces
pfn
PRO
0
140
フロントエンドカンファレンス関西2026 - 速すぎるUIに、待ち時間を
teamlab
PRO
0
210
DGX Spark でどこまでハッキングができるのか?
kash1064
3
1.4k
会社説明資料
persolserverworks
0
9.8k
S3 Filesで挑む! 八戸せんべい汁の食べ頃問題
usanchuu
0
120
データエンジニアの次の戦場は "プロセスエンジニアリング"
sansantech
PRO
1
120
FinOps Agentが生んだ技術負債:コスト自動調査の落とし穴と解消までの道のり
oasis1994liveforever
2
180
CTO Night|NotionのAIエージェントとFDE
kzkhykw1991
0
270
LLM-as-a-Judgeは本当に信頼できるのか――評価器を評価する話
taka_aki
2
790
今さら理解するウェブフォントver.2026
takanorip
3
940
PHPカンファレンス関西の招待状 / Invitation to PHP Conference Kansai
ykanoh
0
220
Featured
See All Featured
The Limits of Empathy - UXLibs8
cassininazir
2
730
Evolution of real-time – Irina Nazarova, EuRuKo, 2024
irinanazarova
9
1.6k
Principles of Awesome APIs and How to Build Them.
keavy
128
18k
Primal Persuasion: How to Engage the Brain for Learning That Lasts
tmiket
0
500
How to train your dragon (web standard)
notwaldorf
96
6.9k
The Myth of the Modular Monolith - Day 2 Keynote - Rails World 2024
eileencodes
28
3.7k
Responsive Adventures: Dirty Tricks From The Dark Corners of Front-End
smashingmag
254
22k
職位にかかわらず全員がリーダーシップを発揮するチーム作り / Building a team where everyone can demonstrate leadership regardless of position
madoxten
69
67k
Claude Code のすすめ
schroneko
67
230k
Between Models and Reality
mayunak
4
500
How to make the Groovebox
asonas
2
2.5k
Kristin Tynski - Automating Marketing Tasks With AI
techseoconnect
PRO
0
550
Transcript
σβΠϯܕ"*ڭҭݚڀηϯλʔཧֶ෦ʢซʣ ӝੜਅ ੜ"*Λ༻͍ͨαʔϏεͷհ .4$PQJMPUɺ(PPHMF(FNJOJɺ1FSQMFYJUZɺ/PUFCPPL-. ੜ"*ϋϯζΦϯηϛφʔ
࣍ جૅతࣄ߲ ֤छ"*αʔϏεͷհ "*αʔϏεͱͷ͖߹͍ํ 1
1 1
جૅతࣄ߲ എܠͷը૾*NBHF'9Ͱੜ
ੜ"*αʔϏεɾπʔϧ ʢ04ʣΈࠐΈ ʢྫʣ"QQMF*OUFMMJHFODF $PQJMPUJO8JOEPXT ͷΞϓϦέʔγϣϯͱͯ͠ར༻ ʢྫʣ.JDSPTPGU&EHF "SD
0MMBNB ΣϒαʔϏεͱͯ͠ར༻ ΣϒαʔϏεͷҰ෦ػೳʹར༻ ʢྫʣ/PJUPO"* 'JHNB"* /BQLJO (BNNB
ੜ"*αʔϏεɾπʔϧ จষɾը૾ͷ࡞ɺ༁ɺ ༩͑ͨใʹجͮ͘ݕࡧɾੜ ը૾ɾಈըͷ࡞ ༩͑ͨใʹجͮ͘ݕࡧɾੜ "*ݕࡧΤϯδϯ ʢྫʣ$IBU(15 (FNJOJ
$MBVEF 1-B.P ରܕੜ"* Σϒϒϥδϯάͱཁ ը૾ੜ"* "*ϊʔτϒοΫ ʢྫʣ/PUFCPPL-. 1FSQMFYJUZ /PUJPO"* ʢྫʣ%SFBN4UVEJP *NBHF'9 #JOH*NBHF$SFBUPS ʢྫʣ.JDSPTPGU&EHF "SD ҰͭͷαʔϏεͰෳͷػೳΛ༗͢Δͷ͋Δ σβΠϯɺ࡞ ʢྫʣ/BQLJO $BOWB 'JHNB (BNNB
ੜ"*αʔϏεͰ͜ΕΒͷϞσϧΛར༻͢Δ ੜ"*Ͱ༻͍ΒΕΔϞσϧʢେنݴޠϞσϧʣ (150QFO"* $MBVEF"OUISPQJD --BNB.FUBʢچ'BDFCPPLʣ (SPLY"* (FNJOJ(PPHMF%FFQ.JOE 6MUSB 1SP 'MBTI
'MBTI# /BOP (155VSCP (15P (15PNJOJ PQSFWJFXͳͲ
ରܕੜ"*ͷൺֱ https://www.nikkei.com/article/DGXZQODL1858X0Y4A910C2000000/ ར༻͢ΔϞσϧʹԠͯ͡ಛಘҙɾෆಘҙ͕͋Δ ͳͥʁ ˠֶशσʔλͷ࣭ͱྔɺֶशϞσϧͷنʢύϥϝʔλʣͳͲ͕ҟͳΔͨΊ
͂ ਓೳΛؚΉػցֶशͷΛѻ͏ ੜ"*ͷར༻ํ๏ ೖྗ <latexit sha1_base64="9l8a28VdIjaXoBKkK/gsygflnqA=">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</latexit> y = f(x) ग़ྗ
ϓϩϯϓτʢࢦࣔจʣ σʔλ ੜ"* ༧ଌɺྨ ೖྗ ग़ྗ Ϟσϧ ίϯςϯπʢจষը૾ɺԻͳͲʣ $IBU(15ͷ͍ํΛڭ͑ͯ ಙౡݝͷΠϝʔδΛը૾Ͱग़ྗͯ͠ ग़ྗͯ͠΄͍͠ใɺ ߦͬͯ΄͍͠࡞ۀΛࢦࣔ͢Δ $IBU(15ͷ͍ํʹ͍ͭͯɺҎԼʹ͍͔ͭ͘ͷϙΠϯτΛ·ͱΊ·͢Ͷɻ ࣭૬ஊΛ͢Δ • ใऩूɿΘ͔Βͳ͍͜ͱΓ͍ͨ͜ͱΛ࣭͢ΔͱɺͰ͖Δ͚ͩΘ͔Γ͓͑͘͢ ͠·͢ɻྫ͑ɺʮ"*ͱԿ͔ڭ͑ͯʯʮ4%(Tʹ͍ͭͯΓ͍ͨʯͳͲɻ • ૬ஊɿࣄษڧɺৗੜ׆ͰͷΈΛ૬ஊ͢Δ͜ͱՄೳͰ͢ɻΈʹରͯ͠ΞυόΠ εαϙʔτΛߦ͍·͢ɻ
ݕࡧ֦ுੜ ༩͑ͨใʹجͮ͘ݕࡧɾੜ 3"(3FUSJFWBM"VHNFOUFE(FOFSBUJPO ඞཁͳͷʁ Կ͕ग़དྷΔʁ --.͕ग़ྗ͢Δใʢ෦ࣝʣɺֶश࣌ͷσʔλʹج͍͍ͮͯΔ ˠ࠷৽ͷใݸผͷঢ়گΛөͨ͠ग़ྗͱͳΒͳ͍Մೳੑ͕͋Δ ֎෦ࣝͱͯ͠Ϣʔβʔ͕৽ͨͳใΛ༩͑Δ ˠ--.͕࣋ͭࣝͷ෦දݱΛิ͏ͨΊʹ֎෦ͷࣝιʔεʹϞσϧΛ ͤ͞Δ͜ͱͰɺ--.͕ੜ͢Δճͷ্࣭͕͢Δ
˞ΞοϓϩʔυˠࣗͰͷར༻ʹݶΒͣɺকདྷతʹֶशʹར༻͞ΕΔՄೳੑ͋Δ ֎෦ࣝͱͳΔॻྨɺσʔλ
ೳͷ࣌ʙ"*Λୈೋɺୈࡾͷͱͯ͠׆༻͢Δʙ */165 065165
֤छ"*αʔϏεͷհ
"*ࡌϒϥβ
.JDSPTPGU$PQJMPU ΞϓϦέʔγϣϯʹࡌ͞ΕΔ.4"*αʔϏεͷ૯শ https://copilot.microsoft.com/ 8PSEɺ&YDFMɺ1PXFS1PJOUɺ5FBNTͷ.4ΞϓϦέʔγϣϯʹࡌ $PQJMPUGPS.JDSPTPGU $PQJMPUJO8JOEPXT νϟοτΛհͨ͠ͷઃఆมߋɾ੍ޚɻΩʔϘʔυ͓ΑͼΞϓϦ͔Βىಈ $PQJMPUJO&EHF "*ࡌͷΣϒϒϥβ ͲΜͳͷؚ͕·ΕΔʁ
ຊ͝հ ࡞ۀ༰ʹԠͨࣗ͡ಈԽݕࡧɺ࡞ۀ࣌ؒͷॖʹޮՌత
&EHF ӾཡதͷΣϒϖʔδʹܝࡌ͞Εͨ༰ʹ͍͍ͭͯ߹ΘͤΔ 1%'Ͱͷจষ .JDSPTPGU&EHFʢΞϓϦέʔγϣϯʣͷΠϯετʔϧ ඞཁͳͷʁ Կ͕ग़དྷΔʁ ΞΧϯτొ https://www.microsoft.com/ja-jp/edge/ "*ࡌͷΣϒϒϥβ
"SD ӾཡதͷΣϒϖʔδʹܝࡌ͞Εͨ༰ʹ͍͍ͭͯ߹ΘͤΔ 1%'Ͱͷจষ "SDʢΞϓϦέʔγϣϯʣͷΠϯετʔϧ ΞΧϯτ࡞ ඞཁͳͷʁ Կ͕ग़དྷΔʁ https://arc.net/gift/4280d3d "*ࡌͷΣϒϒϥβʢ(PPHMF$ISPNFΫϩʔϯʣ
ӾཡதͷΣϒϖʔδͷࣗಈཧ ݕࡧ݁ՌͷϓϨϏϡʔ
&EHF·ͨ"SDΛࢼͦ͏ ʢྫʣཧɾσʔλαΠΤϯεɾ"*ڭҭϓϩάϥϜೝఆ੍ ఆ໘ɿจ෦ՊֶলͷࡦɾऔΓΈʹؔ͢Δௐࠪ https://www.mext.go.jp/a_menu/koutou/suuri_datascience_ai/00001.htm ֓ཁΛ·ͱΊΔʮ͜ͷϖʔδͷ֓ཁʁʯ ೝఆཁ݅ΛΔʮ۩ମతͳೝఆཁ݅Ͳ͜ʹهࡌ͞Ε͍ͯΔʁʯ ͦͷଞʮ1%'ϦϯΫҰཡΛ༻ҙͯ͠ʯ
&EHF·ͨ"SDΛࢼͦ͏
&EHF·ͨ"SDΛࢼͦ͏ ͋Εʁ
&EHF·ͨ"SDΛࢼͦ͏ ϖʔδͷίϯςϯπΛཧղͯ͘͠Εͳ͍ˠઃఆ͔ΒڐՄΛग़͢
&EHF·ͨ"SDΛࢼͦ͏ ࣈນػೳ͕ར༻Մೳͳಈըͷཧ
&EHF·ͨ"SDΛࢼͦ͏ "SDͷ߹ɺ.BYػೳΛ༗ޮԽ͠ɺ ʮ5VSO0O"MMʯ͋Δ͍ݸผͷػೳʹνΣοΫΛ͚ͭΔ "TL0O1BHF
ࣄ࡞ۀͱ࿈ܞ͢Δ"*
ඞཁͳͷʁ (FNJOJ https://gemini.google.com/ (PPHMFͷ֤छαʔϏεͱ࿈ܞͨ͠ݕࡧ ʢ3"(ͷ֎෦ࣝʹϝʔϧɺΧϨϯμʔɺϚοϓɺυΩϡϝϯτΛར༻ʣ Կ͕ग़དྷΔʁ (PPHMFΞΧϯτ ༗ྉ൛͋Γˠ(FNJOJ"EWBODFEʢԁ݄ʣɻΫϥυετϨʔδͷಛయ ը૾ੜʢ*NBHFOʹΑΔɻ$IBU(15ɺ.4$PQJMPUͰՄ
ˠ$PQJMPUGPS.JDSPTPGUతͳࣄ࡞ۀ͕Մೳ (PPHMF%FFQ.JOE͕։ൃ͢Δ--.ʢ(FNJOJʣʹΑΔର (PPHMFαʔϏεʹඥ͘ରܕੜ"*αʔϏε
(FNJOJΛࢼͦ͏ (PPHMFͷ֤छαʔϏεΛݕࡧͷىʹ͢Δ աڈҰؒͰ&$αΠτʢ"NB[POָ ఱʣͰߪೖͨ͠ͷʁ શͰͳ͍ ࠜڌࢿྉʢ͜ͷ߹ϝʔϧʣΛఏࣔ 4FUUJOHTʢઃఆʣˠ&YUFOTJPOT ֦ுػೳ ˠ
(FNJOJΛࢼͦ͏ ʮ!αʔϏε໊ʯΛϓϩϯϓτͷઌ಄ʹ͚ͭͯ
(FNJOJΛࢼͦ͏ จ͔ΒεϥΠυͷ࡞ (FNJOJʢ$IBUʣͰจΛಡ·ͤΔPSυΩϡϝϯτͰݪߘΛಡ·ͤΔ εϥΠυͷϖʔδͱͯ͠ཧͯ͠Β͏ ϓϨθϯςʔγϣϯͰϖʔδΛ࡞
/PUFCPPL-. https://notebooklm.google/ (FNJOJͱϢʔβʔ͔Βͷ֎෦ࣝʹΑΔௐࠪཁɺΦϦδφϧͷϊʔτ࡞ ࢿྉʹجͮ͘ௐࠪʹ༏Εͨ"*Ξγελϯτ ඞཁͳͷʁ Կ͕ग़དྷΔʁ (PPHMFΞΧϯτ ϓϥΠόγʔͷอޢʜ ʮೖྗࢿྉϞσϧ͔ΒͷԠʹ͓͚ΔݸਓσʔλΛ༻͠ͳ͍ʯ ࢿྉதͷจষͷ΄͔ɺਤදʹ͍ͭͯཧղ
ࢿྉʢจষɺ1%'63-ɺԻϑΝΠϧͳͲʣ
/PUFCPPL-.Λࢼͦ͏ ༻ޠͷઆ໌ ιʔεʢ֎෦ࣝʣ ϊʔτʢϝϞʣ ʢྫʣཧɾσʔλαΠΤϯεɾ"*ڭҭϓϩάϥϜೝఆ੍ https://www.mext.go.jp/a_menu/koutou/suuri_datascience_ai/00001.htm ͭͷιʔε.#·Ͱɻ߹ܭ·ͰͷιʔεΛՃՄೳ ϊʔτϒοΫͰϞσϧ͔ΒͷԠΛه ϊʔτϒοΫ /PUFCPPL-.ͰͷϓϩδΣΫτͱͯ͠ѻΘΕΔ
"*ݕࡧΤϯδϯ
1FSQMFYJUZ https://www.perplexity.ai/ (15Λ࢝Ίͱ֤ͨ͠छ--.Λ༻͍ͨΣϒݕࡧɾཁɺϖʔδͷ࡞ ΞΧϯτొෆཁͰར༻Մೳɻͨͩ͠ཤྺΒͳ͍ ༗ྉ൛͋Γˠ1SPʢແྉͰ੍ݶ͖Ͱར༻Մɻυϧ݄·ͨυϧʣ https://www.softbank.jp/mobile/service/perplexity-ai/ ιϑτόϯΫɾϫΠϞόΠϧɾ-*/&.0ͷϢʔβʔͰ͋ΕҰؒ1SP͕ແྉ ඞཁͳͷʁ Կ͕ग़དྷΔʁ ग़యݩͷ໌ه
(15Λج൫ͱͨ͠"*ݕࡧΤϯδϯ ݕࡧ݁ՌΛͱʹʮ1BHFTʯΛ࡞ɺಠࣗυΩϡϝϯτͷՃʹΑΔʮ4QBDFTʯ
(FOTQBSL https://www.genspark.ai/ "*ͷ࡞ۀґཔʢλεΫػೳʣ ɹϑΝΫτνΣοΫʹΑΓϋϧγωʔγϣϯΛ͙ ɹσʔλݕࡧ ඞཁͳͷʁ Կ͕ग़དྷΔʁ ΞΧϯτొෆཁͰར༻Մೳɻͨͩ͠جຊػೳͷΈʹ੍ݶɻཤྺΒͳ͍ (PPHMFΞΧϯτͰΞΧϯτొ ΈࠐΈͷ"*$PQJMPUͱͷର
ϦΞϧλΠϜͳݕࡧͱࠂภΓͷͳ͍ίϯςϯπͷੜ
1FSQMFYJUZ·ͨ(FOTQBSLΛࢼͦ͏ IUUQTXXXQFSQMFYJUZBJTFBSDIOJBOOPIVSPZFRJVSJCFOT 45OLR9%(5),T.F97KNFH ʢྫʣ࠷ۙͷʹ͍ͭͯͷʮ·ͱΊʯ ࠓʢʣͷϓϩٿຊγϦʔζͷ݁Ռ ݄ͷΞϝϦΧେ౷ྖબڍͷ ݄ͷؖࠃେ౷ྖͱࠃձͱͷؒͰىͬͨ͜ࣄ݅
"*αʔϏεͱͷ͖߹͍ํ
"*ͰࣄָʹͳΔͷʁ "*ʮ࡞Γʯʢ"TTJTUJWF*OUFMMJHFODFʣͱͯ͠༗ޮ https://trackingai.org/IQ ʮ"*ɺ͋ΒΏΔ࢈ۀʹ͓͚Δੜ࢈ੑ্ͱΠ ϊϕʔγϣϯʹࢿ͢Δͷ Ͱ͋Γɺੵۃతͳ׆༻Ҏ֎ͷબࢶͳ͍ʯ https://www.keidanren.or.jp/policy/2023/067_honbun.html ʰ"*׆༻ઓུ**ʙΘ͕ࠃͷ"*1PXFSFEԽʹ͚ͯʙʱҰൠࣾஂ๏ਓຊܦࡁஂମ࿈߹ձ "*QPXFSFE
"*QPXFSFE "* "* োน 0OFNPSFUIJOH
"*શੈքͰύϥμΠϜγϑτΛͨΒ͢ ݸਓ͔Β࢈ۀɺֶज़ɺࣾձɺࠃՈɺٿڥ·ͰӨڹΛٴ΅͢ Ϟσϧ։ൃऀΒҰ෦Λআ͍ͯɺ࠷ઌͷ"*ʢϑϩϯςΟΞ"*ʣϞσϧ͕୭Ͱར༻Մೳ ຊͷ߹ɺσδλϧԽͷΕɺʮࣦΘΕͨʯΛṢճ͢Δػձ ిؾɾΨεɾਫಓɾΠϯλʔωοτʹ͙࣍ܦࡁɾࣾձج൫ˠ͓ۚΛ͏Ձ͋Δ ଟ͘ͷαʔϏεɾػೳ͕ແྉͰ͑Δˠࢼ͞ͳ͍खͳ͍
ॾҙ ٵऩ͞ΕͨΓɺผͷαʔϏεͰಉ༷ͷػೳ͕ఏڙ͞ΕΔΑ͏ʹͳͬͨΓͯ͠ ʮαʔϏεऴྃʯ͢Δ ϕʔλ൛ͷؒແྉͰ͑ͨͷ͕ɺ༗ྉػೳʹͳΔ αʔϏε͍ͭ·Ͱ͑ΔͱݶΒͳ͍ ࣍ʑͱ৽αʔϏεػೳ͕ग़ͯ͘Δɻͣͬͱ͍ͬͯΔͱݹ͘ͳΔՄೳੑ ͍Ζ͍Ζͳπʔϧ͕͋Δ͕ɺߜΓͳ͕ΒͬͯΈΔ