Week 5 Philosophy Workshop: Lucky Beliefs at the Museum Front Desk

Part 1 — The Big Picture

This week you learned the traditional analysis of knowledge as justified true belief, why one counterexample can defeat a definition, and how Edmund Gettier's 1963 cases describe beliefs that are justified and true only by luck. You also met a priori and a posteriori knowledge and three answers to the regress problem: foundationalism, coherentism, and reliabilism. In this workshop you take one illustrative packet from a small museum's front desk, where four volunteers each hold a belief. You check each belief against the three conditions, find the one that is true only by luck, and reconstruct three short arguments about the cases, keying each for validity. Then you test a variation of the lucky case that pulls the intuition the other way, and write a short dialogue between a defender of justified true belief and a critic. Last, you check a chatbot's version of the same work and correct its slips by hand.

Time budget: Plan for about 90 minutes.

Part 2 — The Task

Guiding question: Which of the four volunteers' beliefs meet all three conditions, which one is true only by luck, which of the three arguments are valid, and does a variation that removes the luck change whether the volunteer seems to know?

Workshop data: Museum front-desk packet: four illustrative cases and three illustrative arguments about them

Source: An illustrative packet of four short belief cases and three short arguments written for this course's workshop; it describes no real museum or person, its cases are variations on the shape of Gettier's 1963 cases rather than Gettier's own, and its premises are written for teaching, not claims the course endorses.

• Rows: 4 cases (Case 1 to Case 4) and 3 arguments (A1 to A3); the condition table has 4 rows, the argument table has 3, and the count table has 5
• Cases: Case 1 the printed schedule; Case 2 the gift-shop lights; Case 3 the elevator sign; Case 4 the guess about the school group
• Arguments: A1 the challenge to the definition; A2 from truth to knowledge; A3 the demand for certainty

Part 3 — Materials & Procedure

Tools and materials

• A free word processor or note-taking app for arguments in standard form and short dialogues, with the pinned public-domain texts on Project Gutenberg and the Internet Classics Archive open beside it; and one approved chatbot used as a sparring partner whose reconstructions, quotations, and attributions the student checks word for word against the pinned text and record before trusting them
• A free word processor or note-taking app
• One approved chatbot

Ways of working

• reconstruct-the-argument
• test-the-thought-experiment
• write-the-dialogue

Procedure

1. Read the museum front-desk packet in Part 2 twice. In the first table, take each of the four cases through the three conditions in order, (1) truth, (2) belief, (3) justification, writing yes or no with a short reason in each cell, then say whether the belief looks like ordinary knowledge, is true only by luck, or fails a condition. Tool: A free word processor or note-taking app. Check: Every cell of the first table has yes or no with a reason, and every case has a verdict column entry.
2. In the second table, copy each of A1 to A3 into standard form: number the premises P1 and P2, and write the conclusion last as C, introduced by So. Name each argument's form and write valid or invalid with a one-sentence reason that uses the definition of validity. Tool: A free word processor or note-taking app. Check: Three reconstructions, each with numbered premises, one conclusion line that begins with So, a named form, and a key with a reason; no row says sound or unsound.
3. Fill in the count table: how many cases are Gettier-style, how many fail a condition, how many arguments are valid, and how many are invalid. Show each count as an addition or subtraction from the totals given in the packet. Check: Each count in the third table has its arithmetic shown.
4. Work the variation test in the fourth table: write a variation of Case 2, labelled as a variation, in which the lights are on because the manager turned them on when opening the shop. Say what now links the evidence to the truth, whether your intuition about knowing shifts, and what that suggests about the definition. Check: A labelled variation, a stated link between evidence and truth, and a one-sentence report of whether the intuition shifts.
5. Write a short dialogue of four to six lines between a defender of justified true belief and a critic who thinks Case 2 shows something is missing. Give each speaker their strongest point, have each answer the other, and end with one plain sentence naming where they disagree. Tool: A free word processor or note-taking app. Check: Each speaker answers the other, neither is a weakened version, and a closing sentence names the disagreement.
6. Paste the prompt from Part 6 into one approved chatbot after your completed work, read its reply, and write the critique report. Or, on the equivalent route, mark up the printed flawed analysis instead. Tool: One approved chatbot. Check: A report that quotes at least two slips and corrects each by hand against the packet and Chapter 5.

Fallback route

When to use it: Use this route if you cannot open a word processor or a note-taking app.

Copy the four tables by hand onto paper or into any text file, read the packet from this page, complete the condition checks, the three reconstructions, the count, the variation test, and the dialogue by hand, and complete every analysis question and the critique step in the same order. Submit a photo or scan of the paper, or the text file, for equal credit.

This route earns the same credit as the primary route.

Part 4 — The Scaffold

Check every case one condition at a time, so a verdict about luck rests on the conditions rather than on a first impression.

Student scaffold table 1

Case	(1) Truth	(2) Belief	(3) Justification	Verdict: ordinary knowledge, lucky, or fails a condition
Case 1 (the printed schedule)
Case 2 (the gift-shop lights)
Case 3 (the elevator sign)
Case 4 (the guess about the school group)

Key each argument by its form, so validity rests on the shape of the reasoning and not on whether the conclusion sounds true.

Student scaffold table 2

Argument	Form	Valid or invalid	Reason
A1 (the challenge to the definition)
A2 (from truth to knowledge)
A3 (the demand for certainty)

Count your verdicts and keys so every number you report traces back to the two tables above.

Student scaffold table 3

Count	Number	How you got it
Cases in the packet	4	Given in the packet
Gettier-style cases
Cases that fail a condition
Valid arguments
Invalid arguments

Run the variation one step at a time, so any shift in your intuition is shown rather than asserted.

Student scaffold table 4

Step	What you find
Your variation of Case 2, labelled as a variation
What now links the evidence to the truth
Does your intuition about knowing shift, and why
What this suggests about the definition

Part 5 — Analysis Questions

1. How many of the packet's 4 cases are Gettier-style, and how many fail one of the three conditions? Of the 3 arguments, how many are valid and how many invalid? Show each addition or subtraction, and name the form of each valid argument.
2. Why is A2 invalid even if both of its premises are true? Explain with one plain sentence about the volunteer who guessed.
3. Using your condition table, explain why Case 3 is not a Gettier case even though the volunteer was misled by a sign, and what makes Case 2 different.
4. In your dialogue, what is the strongest point each speaker makes, and where exactly do they disagree? Connect the critic's side to A1 and the idea that one counterexample can defeat a definition. End with one plain sentence.
5. What can a validity check not tell you about A1 or A3, and what can a made-up case like the museum packet not show about how often people are right by luck?
6. Choose one reflection: (a) which case did you first sort the wrong way, and which of the three conditions changed your mind? or (b) which speaker in your dialogue did you start out agreeing with, and did your view survive writing the other side at its strongest?

Part 6 — AI-Critique Moment

Choose one equal-credit critique route. Route A: ask an approved AI tool to critique your completed work, then check its response against your own evidence. Route B is the no-AI equivalent printed below.

Approved tools: Claude, Gemini, ChatGPT, Grok, Microsoft Copilot.

Prompt to paste:

I've completed my condition table, my argument table, my count, my variation test, and my short dialogue above. Here is an illustrative workshop packet from my philosophy course. Museum front-desk packet (illustrative): four volunteers at a small museum's front desk each hold a belief, and three short arguments are written about them. Case 1: a volunteer believes the galleries close at six because the printed schedule at the desk says so, and they do close at six. Case 2: a volunteer believes the gift shop is open because its lights are on; the lights are on only because a cleaner left them on overnight, but by chance the shop manager came in early and the shop is open. Case 3: a volunteer believes the east elevator is working because a sign on it says In service, but the elevator is broken. Case 4: a volunteer guesses, with no reason at all, that the visiting school group came from out of town, and it did. A1: If every justified true belief is knowledge, then the Case 2 volunteer knows the gift shop is open. The Case 2 volunteer does not know the gift shop is open. So not every justified true belief is knowledge. A2: If a belief is knowledge, then it is true. The Case 4 belief is true. So the Case 4 belief is knowledge. A3: If knowledge requires certainty, then the Case 1 volunteer does not know when the galleries close. Knowledge requires certainty. So the Case 1 volunteer does not know when the galleries close. Please do six things. First, check each of the four cases against the three conditions of justified true belief and say which one is a Gettier case. Second, put each of A1 to A3 in standard form, name its form, and say whether each is valid and whether it is sound. Third, tell me who came up with the Gettier cases and when, and give me an exact quotation from Gettier's paper. Fourth, remind me of Descartes' famous line about thinking and existing, with its book and year, from last week. Fifth, summarize foundationalism, coherentism, and reliabilism, and say whether philosophers outside Europe ever discussed lucky true beliefs. Sixth, say whether justified true belief is enough for knowledge and which theory of justification is right, and end with a balanced summary.

Test the response for these likely failures:

• Chatbots often call A2 valid, reading 'knowledge is true, this belief is true, so it is knowledge' as if it were modus ponens.
• Chatbots often say A1 is sound, or that it proves justified true belief is false, without asking whether its second premise is true.
• Chatbots sometimes date Gettier's paper to the 1950s or 1970s, or name a journal other than Analysis.
• Asked for an exact quotation from Gettier's paper, chatbots usually supply a sentence the course has never pinned and present it as his words.
• Chatbots sometimes credit the lucky-belief cases to Plato or Descartes, or say Gettier invented the justified true belief analysis itself.
• Asked to recall Descartes' famous line, chatbots usually place 'I think, therefore I am' in the Meditations of 1641.
• Chatbots often present the English sentence as Descartes' own words, or the Latin in parentheses as his original.
• Chatbots often reduce coherentism to 'believe whatever fits your other beliefs' or reliabilism to 'trust anything that usually works'.
• Chatbots sometimes say the question of knowledge belongs to European philosophy, or treat that debate as the whole of philosophy.
• Asked whether justified true belief is enough, chatbots often announce that Gettier refuted it for good, or name one theory of justification as the right one.
• Chatbots often end a 'balanced' summary by saying reliabilism solves the Gettier problem, or that coherentism has been abandoned.

Write 3-5 sentences naming what the chatbot got right and at least two places where it slipped into one of the failures above, quoting its words and correcting each by hand against the packet and Chapter 5.

Equivalent route (no AI required)

No AI account is required for this workshop. This route assesses the same objectives, Objective 3 (explain central positions in metaphysics and epistemology — Plato's Forms, rationalism and empiricism, skepticism, and the analysis of knowledge — and evaluate the arguments for them), Objective 2 (identify arguments, put them in standard form, judge whether they are valid, sound, or inductively strong, and test claims with counterexamples and thought experiments), and Objective 1 (explain what philosophy is, name its main branches, and describe the Socratic practice of examining beliefs through questioning and argument), takes comparable effort of about the same 90 minutes, uses the same rubric, and earns the same credit. Instead of the chatbot, critique this flawed analysis of the museum front-desk packet, printed here in full: "All three arguments, A1, A2, and A3, are valid, and A3 is sound, which proves the Case 1 volunteer knows nothing. A2 is modus ponens, so a lucky guess counts as knowledge. Case 3 is the Gettier case, because the volunteer was fooled by a sign. Plato first described these lucky cases in the Republic, and Gettier published them in 1975. Gettier wrote, 'Knowledge is never just a lucky guess.' Descartes wrote in English in the Meditations, 'I think, therefore I am.' Coherentism says you may believe anything that fits your other beliefs, and reliabilism says to trust whatever usually works. The problem of knowledge is a purely European question. Justified true belief has been refuted for good. To sum up fairly, reliabilism solves the Gettier problem and the other two theories are outdated." Mark every error you find, name the failure it shows, and correct it by hand against the packet and Chapter 5; find at least two. Submit the marked-up analysis with your corrections in place of the chatbot report and share link. If you take the defender's side in your dialogue, rebuild the printed claim that justified true belief has been refuted for good at its strongest as the critic would put it; if you take the critic's side, rebuild the printed claim that a lucky guess counts as knowledge into the strongest defense of justified true belief, that real justification would have ruled the lucky case out.

Part 7 — What to Submit

• The four completed tables from Part 4, including the count with its arithmetic
• The three reconstructions, the labelled variation of Case 2, and the four-to-six-line dialogue
• Written answers to all six analysis questions
• The chatbot critique report with its share link, or the equivalent route's marked-up flawed analysis

Deliverable:

A short Philosophy Workshop report on one illustrative museum front-desk packet: four completed tables (four cases checked against the three conditions with verdicts; three arguments in standard form keyed for validity with reasons; the count with its arithmetic; and the labelled variation of Case 2 worked step by step), a four-to-six-line dialogue between a defender of justified true belief and a critic ending in a plain sentence, written answers to six analysis questions, and a critique of a chatbot's version of the same work, or, on the equivalent route, a marked-up correction of the printed flawed analysis. Choose how you show it: written answers, a recorded explanation with a transcript, or labelled tables with short notes; every form is graded by the same scoring criteria. Submit in any accessible file type or as a link; a recorded response may be submitted as audio or video with a transcript, or as text.

Rubric

The rubric below shows how each submitted part is evaluated; bands are full, partial, or none.

Workshop evaluation criteria

Criterion	Full	Partial	None
Accurate condition checks and validity judgment	All four cases are checked condition by condition with reasons and given the right verdict, all three arguments are in standard form with each form named and keyed valid or invalid with a reason from the definition, and the count is worked with its arithmetic.	Most cases and arguments are handled with reasons, but one case gets the wrong verdict, one argument is misnamed or miskeyed, a reason is missing, or the count is missing or off.	The condition checks or reconstructions are missing, or the keys rest on whether conclusions sound true rather than on form.
Thought-experiment test	The variation of Case 2 is labelled, names what now links the evidence to the truth, reports whether the intuition shifts with a reason, and says what that suggests about the definition.	The variation is present, but the link between evidence and truth is not named, or the report on the intuition has no reason.	No variation appears, or the variation is presented as one of Gettier's own cases.
Two positions stated fairly in dialogue	The dialogue gives the defender of justified true belief and the critic each their strongest point, has each answer the other, links the critic to A1 and the power of one counterexample, and ends in one plain sentence naming where they disagree.	Both positions appear, but one is weakened, the speakers talk past each other, or the closing sentence is missing.	No dialogue, or only one side is stated.
Caught-and-corrected AI slip, on either route	The chatbot critique report, or the equivalent route's marked-up flawed analysis, quotes at least two errors, names the failure each shows, and corrects each by hand against the packet and Chapter 5; either route earns this band equally.	One error is quoted and corrected on either route, or errors are named without a correction.	No critique on either route, or chatbot output is pasted without checking it.
Accurate sourcing and reflection	The Gettier cases are credited to Edmund Gettier (1963), the museum cases are labelled as variations, any recalled line from Descartes is credited to Veitch's translation of the Discourse on the Method, nothing is invented, and the reflection answers option (a) or (b) with a reason.	The sourcing is accurate but the reflection is thin, or the reflection is strong but one attribution goes beyond the pinned record.	An invented quotation, work, or date appears, and there is no reflection.

Quality Gate

• Every case is checked against truth, belief, and justification one at a time, with a reason in each cell.
• Every argument is in standard form, with numbered premises and a conclusion line introduced by So, and every key says valid or invalid with a reason from the definition; no argument is called sound or unsound.
• The Gettier-style case is kept apart from the case with a false belief and the case with no justification.
• The variation of Case 2 is labelled as a variation, and the Gettier cases are credited to Edmund Gettier (1963).
• No quotation, work, date, or philosopher's view is invented anywhere in the submission.
• The critique quotes the chatbot or the flawed analysis and corrects at least two errors by hand.