[{"data":1,"prerenderedAt":219},["ShallowReactive",2],{"topic-page:\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Fgenai-model-selection-and-business-metrics":3},{"topic":4,"category":177,"seo":181,"breadcrumbs":186,"prerequisites":196,"nextTopics":197,"relatedTopics":207},{"id":5,"slug":6,"title":7,"summary":8,"canonicalPath":9,"categoryPath":10,"difficulty":11,"targetAudience":12,"estimatedMinutes":13,"generatedAt":14,"publishedAt":14,"updatedAt":14,"reviewStatus":15,"tags":16,"expectedKnowledge":26,"learningGoal":29,"keyTerms":30,"blocks":89},"it-aws-aws-certified-ai-practitioner-generative-ai-fundamentals-genai-model-selection-and-business-metrics","genai-model-selection-and-business-metrics","生成AIモデルの選択と事業指標","生成AIモデルの選択と事業指標を、AIF-C01の公式objective 2.2.3・2.2.4に沿って、用語、判断順序、例、誤解修正で整理します。","\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Fgenai-model-selection-and-business-metrics","\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals","beginner","cert_candidate",6,"2026-08-06","ai_generated",[17,18,19,20,21,22,23,24,25],"AWS Certified AI Practitioner","AIF-C01","生成AIの基礎","モデル選択","性能要件","レイテンシ","ROI","変換率","顧客生涯価値",[27,28],"AWSクラウドの基本用語を聞いたことがある","AIを使う業務例を日常的な言葉で考えられる","生成AIモデルの選択と事業指標について、試験で問われる違いと選択基準を説明し、短い業務例へ適用できる。",[31,38,42,45,49,53,57,63,67,70,74,78,85],{"id":32,"label":20,"shortDefinition":33,"supplement":34,"role":37},"model-selection","用途の能力・制約・費用・統制条件に合うモデルを比較して選ぶ活動です。",{"kind":35,"text":36},"misconception","最大のモデルが、すべての用途で最適とは限りません。","primary",{"id":39,"label":40,"shortDefinition":41,"role":37},"modality","モダリティ","文章・画像・音声・動画など、モデルが扱うデータ形式の種類です。",{"id":43,"label":22,"shortDefinition":44,"role":37},"latency","要求を送ってから結果が返るまでにかかる時間です。",{"id":46,"label":47,"shortDefinition":48,"role":37},"model-complexity","モデルの複雑さ","モデルの規模・構造・計算量など、運用負担へ影響する複雑さです。",{"id":50,"label":51,"shortDefinition":52,"role":37},"compliance","コンプライアンス","法令・規制・契約・社内規程などの要求へ適合することです。",{"id":54,"label":55,"shortDefinition":56,"role":37},"business-value","事業価値","技術の利用が、収益・費用・時間・品質などの業務成果へ生む価値です。",{"id":58,"label":23,"shortDefinition":59,"supplement":60,"role":37},"roi","投資に対して得られた利益や効果の大きさを表す指標です。",{"kind":61,"text":62},"usage","導入費・運用費・確認費用も含めて評価します。",{"id":64,"label":65,"shortDefinition":66,"role":37},"efficiency","効率性","同じ成果を、より少ない時間・費用・作業量で得られる程度です。",{"id":68,"label":24,"shortDefinition":69,"role":37},"conversion-rate","利用者のうち、購入や申込など目的行動へ進んだ割合です。",{"id":71,"label":72,"shortDefinition":73,"role":37},"average-revenue-per-user","利用者当たり平均売上（ARPU）","一定期間の売上を利用者数で割った平均額です。",{"id":75,"label":76,"shortDefinition":77,"role":37},"customer-lifetime-value","顧客生涯価値（CLV）","一人の顧客が関係期間を通じて生む価値の見積もりです。",{"id":79,"label":80,"shortDefinition":81,"supplement":82,"role":84},"accuracy","正解率","全予測のうち、正しく判定できた割合です。",{"kind":35,"text":83},"不均衡データでは、高い正解率だけで目的達成を判断できません。","contextual",{"id":86,"label":87,"shortDefinition":88,"role":84},"context-window","コンテキストウィンドウ","モデルが一回の処理で参照できるトークン範囲です。",[90,95,106,115,124,130,136,142,158,167],{"id":91,"type":92,"title":7,"subtitle":93,"shortDefinition":94},"hero","HeroBlock","能力・制約・費用を、事業成果の測定へつなぐ","モダリティ、品質、速度、入力長、カスタマイズ、法令・社内統制、費用などを、用途の成功条件と照合する評価軸です。",{"id":96,"type":97,"term":20,"definition":98,"plainExplanation":99,"presentation":100,"supportLabel":101,"keywords":102},"definition","DefinitionBlock","[[cloze:model-selection|モデル選択]]とは、用途の能力・制約・費用・統制条件に合うモデルを比較して選ぶ活動です。","[[term:modality|モダリティ]]、品質、[[term:latency|レイテンシ]]、[[term:context-window|入力長]]、[[term:model-complexity|モデルの複雑さ]]、[[term:compliance|コンプライアンス]]、費用を用途の必須条件と照合します。","natural","見分けるポイント",[103,104,105],"必須条件","同条件比較","事業成果",{"id":107,"type":108,"steps":109,"title":114},"decision-steps","StepBlock",[110,111,112,113],"業務成果と最低品質を測定可能な形で定義する。","必要モダリティ、言語、入力長、速度、統制条件で候補を絞る。","代表データで品質・遅延・単位費用を同じ条件で比較する。","本番指標を基準値と比較し、継続・変更・停止を判断する。","選択から価値検証まで",{"id":116,"type":117,"title":118,"points":119,"body":123},"exam-cues","KeyPointBlock","判断で押さえる境界",[120,121,122],"用途の必須条件で候補を絞る","代表データで同条件比較する","モデル指標を事業成果へ接続する","分類用途なら[[term:accuracy|正解率]]などの技術指標を測り、[[term:efficiency|効率性]]、[[term:conversion-rate|変換率]]、[[term:average-revenue-per-user|利用者当たり平均売上（ARPU）]]、[[term:customer-lifetime-value|顧客生涯価値（CLV）]]、[[term:roi|ROI]]へ因果の鎖を作ります。指標が改善しても返品や苦情が増えるなら[[term:business-value|事業価値]]とは言えず、品質・費用・顧客影響を組にして監視します。",{"id":125,"type":126,"src":127,"alt":128,"caption":129},"diagram","DiagramBlock","\u002Fassets\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Fgenai-model-selection-and-business-metrics\u002Fgenai-model-selection-and-business-metrics.svg","生成AIモデルの選択と事業指標について能力・モダリティ、品質・遅延・費用、業務効率・変換率、ROI・顧客価値の関係を文字ラベルと矢印で示した図","モデルの技術特性を、業務指標、最終的な事業成果へつなぐ四層図です。上位成果だけをモデルへ直接帰属させません。",{"id":131,"type":132,"scenario":133,"explanation":134,"result":135},"worked-example","ExampleBlock","商品説明文を生成するモデルSとLを比較します。Lは品質が少し高い一方、費用と遅延が大きい状況です。","代表商品で事実誤り、ブランド表現、生成時間、担当者の修正時間、一件費用を測ります。Lの品質差が修正時間や変換率を十分改善しないならSが合理的です。逆に多言語や長い仕様書でLだけが最低品質を満たすなら、その対象だけLを使う分岐もできます。","単一の『最高モデル』ではなく、対象と価値に合うモデル構成を選べます。",{"id":35,"type":137,"warningTitle":138,"message":139,"severity":140,"action":141},"WarningBlock","ROIは売上増だけでは測れない","導入費、推論費、監視、担当者確認、誤生成への対応も費用です。また売上変化には他施策も影響するため、比較期間や対照を設計します。","warning","各技術指標が、どの業務行動を通じて事業指標へ影響するかを書きます。",{"id":143,"type":144,"title":145,"questions":146},"quiz","QuizBlock","AIF-C01 判断チェック",[147],{"question":148,"choices":149,"answerIndex":153,"choiceExplanations":154},"モデル候補を公平に比較する方法はどれですか。",[150,151,152],"同じ代表データと成功基準で品質・遅延・費用を測る","モデル名の知名度だけで決める","最大パラメータ数だけを事業価値とみなす",0,[155,156,157],"正解です。用途に沿った同条件比較です。","要件適合を確認できません。","規模だけでは品質やROIは決まりません。",{"id":159,"type":160,"title":161,"points":162},"summary","SummaryBlock","まとめ",[163,164,165,166],"能力・制約・統制・費用でモデルを選ぶ","代表データと共通基準で比較する","技術指標を業務指標へつなぐ","ROIには継続運用と誤り対応の費用も含める",{"id":168,"type":169,"title":170,"nextTopics":171,"relatedKeywords":173},"next-action","NextActionBlock","次は「AWS生成AIサービス・基盤・費用トレードオフ」へ",[172],"it-aws-aws-certified-ai-practitioner-generative-ai-fundamentals-aws-genai-services-infrastructure-and-cost-tradeoffs",[174,175,176],"Amazon Bedrock","SageMaker AI","Amazon Q",{"path":10,"title":19,"description":178,"parentPath":179,"accentToken":180},"生成AIの構成概念、ユースケース、エージェント型AI、モデル選択とAWS基盤をAIF-C01の範囲で整理します。","\u002Fit\u002Faws\u002Faws-certified-ai-practitioner","cloud-amber",{"title":182,"description":183,"canonicalUrl":184,"ogImage":185},"生成AIモデルの選択と事業指標 | AWS Certified AI Practitioner","AIF-C01対策として生成AIモデルの選択と事業指標の定義、判断軸、具体例、誤解しやすい境界を公式試験範囲に沿って学びます。","https:\u002F\u002Fzeqnilo.com\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Fgenai-model-selection-and-business-metrics","https:\u002F\u002Fzeqnilo.com\u002Fassets\u002Fsite\u002Fdefault-og.svg",[187,190,193,194,195],{"title":188,"path":189},"IT","\u002Fit",{"title":191,"path":192},"AWS","\u002Fit\u002Faws",{"title":17,"path":179},{"title":19,"path":10},{"title":7,"path":9},[],[198],{"id":172,"title":199,"summary":200,"canonicalPath":201,"categoryPath":10,"categoryTitle":19,"difficulty":11,"targetAudience":12,"estimatedMinutes":13,"publishedAt":14,"updatedAt":14,"reviewStatus":15,"tags":202},"AWS生成AIサービス・基盤・費用トレードオフ","AWS生成AIサービス・基盤・費用トレードオフを、AIF-C01の公式objective 2.3.1・2.3.2・2.3.3・2.3.4に沿って、用語、判断順序、例、誤解修正で整理します。","\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Faws-genai-services-infrastructure-and-cost-tradeoffs",[17,18,19,174,175,176,203,204,205,206],"Kiro","Strands Agents","Bedrock AgentCore","生成AI費用",[208],{"id":209,"title":210,"summary":211,"canonicalPath":212,"categoryPath":10,"categoryTitle":19,"difficulty":11,"targetAudience":12,"estimatedMinutes":13,"publishedAt":14,"updatedAt":14,"reviewStatus":15,"tags":213},"it-aws-aws-certified-ai-practitioner-generative-ai-fundamentals-genai-strengths-limitations-and-risk","生成AIの強み・限界・リスク","生成AIの強み・限界・リスクを、AIF-C01の公式objective 2.2.1・2.2.2に沿って、用語、判断順序、例、誤解修正で整理します。","\u002Fit\u002Faws\u002Faws-certified-ai-practitioner\u002Fgenerative-ai-fundamentals\u002Fgenai-strengths-limitations-and-risk",[17,18,19,214,215,216,217,218],"生成AI","幻覚","非決定性","解釈性","人による確認",1787370540079]