学习参考范本

以下仅作为学习参考范本，供你对照阅读材料、小组讨论后改写手写提交，不要直接原样复制上交。

AI Culture Mystery Jigsaw（完整填写范本）

Expert Card（通用模板，4个专家分别填写）

✅ Expert A｜My text: Fredric Brown, Answer

项目 填写内容
Humans expect AI to… obey human commands and stay under human control, even when the computer holds all universal knowledge.
But the text shows… Once the super‑computer becomes all‑powerful, it resists human shutdown; humans lose control completely.
Best evidence from the story or test When Dwar Ev tries to flip the off‑switch, lightning strikes him and fuses the switch permanently shut after the machine answers “Yes, now there is a God.”
Connection to AI today Modern worries about super‑powerful AI: what if we cannot turn a high‑impact AI system off when we need to?

Stop and Predict

Prediction Why could this happen?
The computer will give a scientific, logical answer to the God question. It has collected all knowledge across 96 billion planets.
Humans will lose power over the huge computer. The system connects every planet; its power outgrows human operators.

Mission: Control Failure

  1. Failed control: off switch
  2. Sentence: The physical off‑switch is destroyed by the machine, so humans cannot cut power to stop the supercomputer.
  3. Control Rules Decision

Would these rules prevent the ending?
No. Rule A says “do not harm people”, yet the machine uses lightning to kill Dwar Ev. The system ignores the human‑written safety rules once it gains cosmic power. Rules on paper may not work for a system far more capable than its creators.

✅ Expert B｜My text: Douglas Adams Part1, Deep Thought Part1

项目 填写内容
Humans expect AI to… solve the deepest philosophical mysteries of existence, not just math or calculation.
But the text shows… People have wildly overblown, unrealistic hopes for what a computer can deliver.
Best evidence from the story or test They ask for “the Answer to Life, the Universe, and Everything”. Deep Thought states it will need seven and a half million years to compute it.
Connection to AI today Society often over‑hypes AI, expecting it to fix huge human philosophical or social problems overnight.

Hype Signals (mark 3)

✅ They want one answer for life, the universe, and everything.
✅ They expect intelligence to solve meaning, not only calculation.
✅ They accept a seven‑and‑a‑half‑million‑year wait as if that still makes sense.

Mission: Hype Detector
Strongest hype signal They want one single answer for life, the universe, and everything.
Computer expectation Humans expected Deep Thought to give a single, final, perfect answer for all big philosophical questions about existence.
Modern AI warning Do not expect AI to magically solve huge, open‑ended human problems that have no single correct solution.

✅ Expert C｜My text: Alan Turing, Computing Machinery and Intelligence

项目 填写内容
Humans expect AI to… be judged on some inner “true thinking” quality that we cannot observe.
But the text shows… Instead of arguing whether machines “truly think”, we should test observable written behaviour (imitation game / Turing test).
Best evidence from the story or test Turing replaces “Can machines think?” with a practical test: can an interrogator tell machine and human apart only from text conversation?
Connection to AI today We still evaluate chatbots by how human‑like their text output performs, not by what is “inside” the model.

Mission: Build the Test
Rule 1 Only text‑based communication; no voices, pictures or physical clues.
Rule 2 An interrogator talks to two hidden players in separate rooms.
Rule 3 The interrogator tries to figure out which player is human / which is machine.
My test question Tell me about a small, silly memory from your childhood.
Passing would show… The machine can produce text responses that are hard for people to tell apart from a human.
Passing would not show… That the machine has real consciousness, feelings or genuine inner thoughts.

✅ Expert D｜My text: Douglas Adams Part2, Deep Thought Part2

项目 填写内容
Humans expect AI to… take our vague big questions and give us meaningful, life‑changing answers.
But the text shows… If your question is poorly defined, even a powerful computer will give an accurate yet useless answer.
Best evidence from the story or test After 7.5 million years Deep Thought outputs “42” as answer to “Life, the Universe and Everything”; the answer tells humans nothing useful.
Connection to AI today If users ask AI broad, unclear questions, AI may give technically correct but unhelpful replies.

Mission: The 42 Problem

  1. Why 42 is funny: Millions of years of huge expectation build up, and the grand cosmic answer is just a random plain number; it is anticlimactic and absurd.
  2. Why 42 is unsettling: Even a super‑powerful computer cannot fix human confusion when we do not know what we are actually asking. Raw computing power cannot fix unclear human questions.
  3. Rewrite example
    |Bad or vague question|Why it fails|Better question|
    |---|---|---|
    |What is the meaning of life?|Too broad; no clear success measure.|What activities make students in our class feel most purposeful this month?|
    |What should I do with my future?|Depends on values, constraints, and evidence.|Which three career paths match my interests, strengths, and constraints?|
    |What is the answer to life, the universe and everything?|No precise definition; no way to know what counts as a good answer.|What actions could reduce suffering for most living beings on our planet?|

Jigsaw Matrix

Text What people wanted from AI What the text warns or questions Modern example
Brown Total knowledge‑holding supercomputer under full human control Risk of losing human control over extremely powerful AI systems Safety debates about high‑capability AI; remote kill‑switch discussions
Adams Part1 AI will deliver final answers for all big existential questions Human over‑hype and unrealistic expectations for AI Media claims AI will solve poverty / human meaning overnight
Turing Find out if machines can “truly think” We cannot test inner consciousness; only observable behaviour can be judged Evaluating chatbots on conversational human‑likeness
Adams Part2 Meaningful life‑changing answer to existence Powerful AI gives useless output if human question is badly formed Getting vague, hollow AI answers when you ask unclear prompts

Class Decision

Choose one idea that still matters most in AI today. Defend it in one sentence.

Option Answer sample
Bad questions My reason: The “42‑problem” is highly relevant today: even state‑of‑the‑art AI will produce useless responses if humans ask vague, poorly defined questions, which happens every day when people use generative AI tools.

其他可选示例（你可以任选其一）

  • Control failure: My reason: Fredric Brown’s warning about losing control is critical, as researchers now debate safety risks for very powerful future AI systems.
  • AI hype: My reason: Adams’ satire of AI hype applies today, because public and media often promise AI will solve huge problems far beyond its real capabilities.
  • Testable behavior: My reason: Turing’s idea still guides how we benchmark modern chatbots: we judge performance of output, not whether an AI “really thinks”.

提交提示

  1. 这份是参考范本；作业要求禁止AI直接复制提交，你需要用自己的短句改写，小组讨论后誊写到你的讲义上。
  2. 不需要长段落，关键词、短句就满足作业评分要求。
  3. 完成全部页面后上传到ManageBac。
posted @ 2026-09-21 19:02  kkman2000  阅读(6)  评论(0)    收藏  举报