シャドーイング練習: AI Agents vs Business Rules: Which Should Make Decisions? - 動画で英語スピーキングを学ぶ
レッスンを作成中...
1
Both business rules and AI agents are ways to automate a decision.
2
So maybe that's to approve a refund or maybe to flag a transaction, but they go about this in different ways.
3
So my question to you is this.
4
Have AI agents superseded business rules or do they both have their place?
5
And what's the difference between these things anyway?
6
Well, let's start with business rules.
7
And business rules run in a business rules engine.
8
That has logic that a person wrote down explicitly, a person like me.
9
So here's a business rule.
10
We could say, if we have an order that is less than 30 days old, and we also have a condition here, which is the item.
11
Was not marked as a final sale.
12
So if and, and we've got a then, the then is we would actually approve the refund.
13
That is a business rule and that rule has conditions and an action.
14
Now an AI agent doesn't necessarily work from rules that are written down.
15
It works a bit differently because with an AI agent you give it a goal and you give it some context of the situation and you also give it access to some tools that it's allowed to call.
16
And the agent works out the steps itself.
17
So instead of providing rules for a refund it's more like hey here's the customer and here's the order policy and here's the customer's order history.
18
Now you sort it out Mr. AI agent, it's up to you.
19
So which one should be making the decision?
20
Well, one thing about a rules engine is you get a consistent and predictable answer.
21
It checks against conditions, against facts.
22
So if the fact is that an order let's say is 12 days old, that's a fact.
23
And if there is a fact that the order was not marked as a final sale, well then the condition comes from the rule.
24
It's under 30 days, it was not a final sale, both are true.
25
So the rule resolves as approve this condition, approve the refund.
26
There's not any guesswork here because it's simple boolean logic.
27
The order is either under 30 days or it is not and that's what deterministic means, the output is a fixed function of the input and business rules are deterministic.
28
Now AI agents don't work like this because they are built with large language models and we know that a large language model predicts next tokens.
29
So given the order and the policy and the goal, the model works out a response from patterns that it's picked up in training and it's doesn't necessarily fix on one final answer.
30
It's working from a probability distribution over possible responses and picking from it.
31
So for something like this, approve is still the most likely outcome, but deny as an outcome still has some probability as well.
32
And that makes AI agents probabilistic.
33
Whereas the rule engines evaluates a fixed condition.
34
The agent picks from a range of likely answers, which is why you can run the same request twice and get different calls.
35
So that's the split.
36
Rules engine, deterministic, the agent, probabilistic.
37
So when to use each one?
38
Well, business rules are used when the decision is well-defined, meaning you can define the logic ahead of time.
39
So the refund policies are pretty good example of that.
40
But then so is something like a loan eligibility check or an insurance pricing tier.
41
Anything where the inputs are structured and you already know what the answer should be.
42
And then in regulated work, there's another reason as well.
43
The rule that was processed is by its very definition, also the explanation.
44
A rules engine points at the exact condition that triggered the decision.
45
So that is great for audits.
46
And there's also determinism which helps with testing.
47
Each branch can be unit tested and it will behave the same way in production.
48
It also runs without a model call so there is no AI inference cost.
49
So there's lots of advantages here to business rules.
50
Now AI agents are for when the logic cannot be written down ahead of time because the result all depends upon understanding the context of a situation.
51
Now the input can be messy or it can be unstructured or maybe there's just a case that nobody anticipated.
52
So should a refund actually be approved?
53
If it turns out I sent back my order of dumbbells for being too heavy, which is arguably the point of dumbbells.
54
That sort of thing is the wheelhouse of AI agents judgment, judging the merits of a reason for a return.
55
That's not really something that's easily captured in deterministic pre-defined rules.
56
So look, that's the the trade-off.
57
A rules engine only handles what was anticipated.
58
It will struggle to handle an input that nobody has a rule for, whereas an agent will generalize because it's working from patterns instead of a fixed branch.
59
So back to the question we started with.
60
Have AI agents superseded business rules?
61
Well, in a lot of real systems, the answer is actually to run the two together.
62
It's just a hybrid approach where you end up using both.
63
So let's fill in this flow with a hybrid solution so we can see what it might look like.
64
So at the top here we start with the request which has come In.
65
And the first thing in the process flow is we're going to navigate to the rules engine.
66
So we'll come down to here.
67
Now why rules first and not an agent first?
68
Well, because rules are comparatively cheap and they're quick to run.
69
So the rules engines checks the request against the policy and then it will determine an action, basically what we should do.
70
So the order is 12 days old, it's not a final sale.
71
That would mean it's under the auto approve limit.
72
And we are going to go down here to the final decision.
73
And that final decision will be to approve that request.
74
That's nice and easy.
75
And if something is clearly outside of the policy, then instead of approving that request, we would decline the request.
76
But that's the decision right there.
77
We're basically done.
78
And most refunds and decisions are gonna be clear cut like this.
79
It's only when the rules can't decide that the request instead will go down this pass where it will be escalated and it's escalated to an AI agent.
80
And there's usually a couple of reasons for that kind of escalation.
81
Now the first is that the data itself is going to need some additional work.
82
So a rule needs clean structured facts to check like an order age or a final sale flag, but plenty of refund requests they're going to turn up with data that's not all that clean.
83
They're going have a bunch of unstructured input like a a paragraph of free text or maybe a picture.
84
Rules engines they by and large work best with structured data, and the agent can also make use of tools as well.
85
Now it might call a vision model to look at that photo of a defective blender or to to query the order system for the customer's purchase history for example, but the agent will work out which tools the situation calls for and in which order those tools should be called and it does this without being scripted in advance.
86
So at that point an agent can recommend a decision of its own, but it might not get the final word because its recommendation is then going to go down to a set of deterministic guardrails where it will then be evaluated.
87
So a guardrail might specify where anything the agent green lights above a certain threshold should actually route somewhere else.
88
It should go first of all to this box here which is human in the loop before the final decision is ultimately reached.
89
It's under a certain threshold it can go directly to the decision.
90
So we have the potential to check the agent's judgment with a real person in situations where the stakes are highest.
91
So no, agents have not superseded hard-coded business rules, but they can work pretty well with them particularly when the data is messy or when we need judgment or where tool calling can expand the context.
92
Ultimately, the two together can help make better decisions.
📺 同じチャンネル
C12
What is Event Driven Architecture (EDA)?
Business EnglishUniversity StudentsWorking Adults
C12
What Is Jev? The AI Model That Doesn't Generate Text
Business EnglishUniversity StudentsWorking Adults
C18
AI, Machine Learning, Deep Learning and Generative AI Explained
Business EnglishUniversity StudentsWorking Adults
C11
Spec-Driven Development: AI Assisted Coding Explained
Business EnglishUniversity StudentsWorking Adults
C13
AI & Education: Generative AI & the Future of Critical Thinking
Business EnglishUniversity StudentsWorking Adults
C12
EDR vs. EPP vs. NGAV
Business EnglishWorking Adults
C13
MCP vs ADK: How Modern AI Agents Connect and Work Together
University StudentsWorking Adults
C11
Cybersecurity Architecture: Data Security
Business EnglishUniversity StudentsWorking Adults
C113
The 7 Skills You Need to Build AI Agents
Business EnglishUniversity StudentsWorking Adults
C118
How Smart Organizations Will Use AI: Jevons Paradox & Workforce Impact
Business EnglishUniversity StudentsWorking Adults
C111
What are AI Agents?
Business EnglishUniversity StudentsWorking Adults
C16
Cybersecurity Architecture: Five Principles to Follow (and One to Avoid)
Business EnglishUniversity StudentsWorking Adults
C11
Zero Trust Explained in 4 mins
Business EnglishUniversity StudentsWorking Adults
✨ おすすめ動画
C11
105 文
AI Governance Explained: Ethical, Transparent & Accountable AI for the Future
Business EnglishUniversity StudentsWorking Adults
C11
635 文
Decision-Making and Entrepreneurial Performance: Evidence from a Field Experiment
Business EnglishUniversity StudentsWorking Adults
C14
321 文
8 Tech Choices to Lock In Before Agentmaxxing
Business EnglishUniversity StudentsWorking Adults
C11
125 文
Introduction to statistical prediction models
Business EnglishUniversity StudentsWorking Adults
C11
149 文
Inside a warehouse run by AI robots
Business EnglishUniversity StudentsWorking Adults
C12
64 文
P2 什么是机器学习【2024公认最好的 | 吴恩达机器学习 | 教程 | Machine Learning Specialization(超爽中英!)】
Business EnglishUniversity StudentsWorking Adults
B21
267 文
ALL OF BUSINESS Explained in 20 Minutes
Business EnglishUniversity StudentsWorking Adults
B228
26 文
Advantages and Disadvantages of AI
IELTS PrepBusiness EnglishUniversity Students
C143
6 文
Ray Dalio on the AI Bubble
Business EnglishUniversity StudentsWorking Adults
C13
10 文
The Data-Driven Decision Framework Every Marketer Needs
Business EnglishUniversity StudentsWorking Adults
このレッスンの語彙とスピーキングのポイント
この動画には、シャドーイング用の文が92文、単語が1411語あります。 音声の長さは10:12です。 話す速さは1分あたり約138語で安定しており、シャドーイングしやすいペースです。 単語の86%は英語の頻出3,000語に含まれます。残りは練習の前に確認しておきましょう。
この動画の重要語彙
動画に出てくるやや難しい単語15語を、発音と意味つきで紹介します。
- approve /əˈpɹuːv/ (動詞) — 承認する, 賛成する. To officially sanction; to ratify; to confirm; to set as satisfactory.
- refund /ɹɪˈfʌnd/ (名詞) — 払い戻し, 返金. An amount of money returned.
- judgment /ˈd͡ʒʌd͡ʒ.mənt/ (名詞) — 判決, 裁き. The act of judging.
- hybrid /ˈhaɪ.bɹɪd/ (名詞) — 雑種. Offspring resulting from cross-breeding different entities, e.g. two different species or two purebred parent strains.
- threshold /ˈθɹɛʃ(h)oʊld/ (名詞) — 敷居. The lowermost part of a doorway that one crosses to enter; a sill.
- probability /ˌpɹɑ.bəˈbɪl.ə.ti/ (名詞) — 確率. The state of being probable.
- evaluate /ɪˈvaljʊeɪt/ (動詞) — 評する. To draw conclusions from examining; to assess; to appraise.
- messy /ˈmɛsi/ (形容詞) — 乱雑な, きたない. In a disorderly state; chaotic; disorderly.
- anticipate /ænˈtɪs.ɪ.peɪt/ (動詞) — 予想する. To act before (someone), especially to prevent an action.
- explanation /ˌɛkspləˈneɪʃən/ (名詞) — 説明. The act or process of explaining.
- output /ˈaʊtpʊt/ (名詞) — 出力. Production; quantity produced, created, or completed.
- capture /ˈkæp.(t)ʃɚ/ (動詞) — 捕獲する. To take control of; to seize by force or stratagem.
- script /skɹɪpt/ (名詞) — 文書, 文章. A writing; a written document.
- expand /ɪkˈspænd/ (動詞) — 広げる, 拡大する. To change (something) from a smaller form or size to a larger one; to spread out or lay open.
- deny /dəˈnaɪ/ (動詞) — 否む, 打ち消す. To disallow or reject.
動画に出てくる句動詞
- write down (動詞) — 書き留める, 記す. To produce or set (something) down in writing; to record something.
- go down (動詞) — 降りる, 降下する. To descend; to move from a higher place to a lower one.
注意したい発音
話し手はthey're, doesn't, we'reなど、短縮形や弱形を12回使っています。聞こえたとおりの短い形で発音しましょう。
- 「sh」と「zh」の音: threshold /ˈθɹɛʃ(h)oʊld/, explanation /ˌɛkspləˈneɪʃən/, capture /ˈkæp.(t)ʃɚ/, transaction /tɹænˈzækʃən/, recommendation /ˌɹɛkəmɛnˈdeɪʃən/
- 長い単語（アクセントの位置に注意）: necessarily /ˌnɛs.əˈsɛɹ.ə.li/, probability /ˌpɹɑ.bəˈbɪl.ə.ti/, anticipate /ænˈtɪs.ɪ.peɪt/, explanation /ˌɛkspləˈneɪʃən/, differently /ˈdɪf.ə.ɹənt.li/
この動画での練習方法
- まず声を出さずに動画を最後まで聞き、知らない単語をメモします。
- 通常の速度で一文ずつシャドーイングし、話し手のリズムに合うまで繰り返します。
- 自分の声を録音して元の音声と比べます。approve, refund, judgmentなどの単語に特に注意しましょう。
この動画の文法
話し手がよく使っている文型を、動画の実際の表現とともに紹介します。
| 文型 | 動画での表現 |
|---|---|
| 受動態 be + 過去分詞 — 誰がするかより、何が起きるかに焦点を当てる | Was not marked · are written · are built |
| 関係詞節 who / which + 節 — 人や物について情報を加える | here, which is · answers, which is · heavy, which is |
| 現在完了形 have/has + 過去分詞 — 過去の出来事が今も関係している | has come · have not superseded |
シャドーイングとは？英語上達に効果的な理由
シャドーイング（Shadowing）は、もともとプロの通訳者養成プログラムで開発された言語学習法で、多言語習得者として知られるDr. Alexander Arguelles によって広く普及されました。方法はシンプルですが非常に効果的：ネイティブスピーカーの英語を聞きながら、1〜2秒の遅延で声に出してすぐに繰り返す——まるで「影（shadow）」のように話者を追いかけます。文法ドリルや受動的なリスニングと異なり、シャドーイングは脳と口の筋肉が同時にリアルタイムで英語を処理・再現することを強制します。研究により、発音精度、抑揚、リズム、連音、リスニング力、そして会話の流暢さが大幅に向上することが確認されています。IELTSスピーキング対策や自然な英語コミュニケーションを目指す方に特におすすめです。