Speaker Notes
Walk through what is visible on each slide · ~1 min per slide
Title Slide
Left panel (burgundy): label ‘UX Research’, title lines ‘Identity Disclosure’ / ‘in Service Chatbots’, tagline ‘Trust · Customer Retention · UX Design’.
Right panel: Subject — User Experience Design · Duration — 6 minutes · 6 slides · Date — June 2026.
Say: This presentation examines how chatbots should disclose their non-human identity in service contexts.
Slide 1 — Why Identity Disclosure Matters
Header: ‘Why Identity Disclosure Matters’ · subtitle ‘Problem · Research question · Relevance’ · top right ‘Thesis Presentation’.
Centre box — ‘Central Research Question’: read the question on the slide verbatim: What role does disclosing a chatbot's non-human identity play in trust and customer retention?
Four bullets below (left to right): (1) service chats — users increasingly interact with chatbots; (2) trust and positive experiences essential for churn prevention; (3) AI hospitality market ~$3.2B by 2025; (4) transparency about machine identity as a key UX factor.
Footer slide number: 1 »
Slide 2 — Theoretical Background
Header: ‘Theoretical Background’ · subtitle ‘Trust as a driver of customer retention’.
Left — paper card: ‘Funke et al. (2023)’. Below it the 2×2 matrix on screen: columns ‘Disclosed AI’ / ‘Concealed AI’, rows ‘Good outcome’ / ‘Bad outcome’. Highlight bottom-right cell ‘Lowest trust’ (concealed AI + bad outcome).
Right sidebar — ‘Key takeaways’: stat ‘>80% preferred disclosed AI’; three bullets on slide — disclosure × outcome shape trust; concealed AI + bad outcome → lowest trust; our study extends Funke's 2×2 design.
Footer: 2 »
Slide 3 — Methodological Approach
Header: ‘Methodological Approach’ · subtitle ‘Online vignette study · n = 37 · randomized order’.
Rationale line (burgundy bar): ‘Prior work: disclosed vs. concealed AI — we split disclosure to test how explicit it should be.’ Explain: literature only had disclose vs. disguise; we wanted to know what disclosure actually looks like.
Three vignette screenshots — note under A/B labels on slide:
A — ‘Radically transparent’ · ‘AI stated explicitly in chat’
B — ‘Subtle disclosure’ · ‘AI powered label only · not in chat’
C — ‘Human disguise’ (concealed-AI baseline, no extra note)
Footer text: design from Funke et al. · measures on 7-point Likert.
Footer: 3 »
Slide 4 — Results
Chart with mean scores — note the animated bracket on Empathy between A and C: p = .017* (only significant paired test).
Key finding box below: descriptive leaders + t-test summary in one sentence.
Small footnote line: paired t-tests, n = 35, exploratory.
Footer: 4 »
Slide 5 — Discussion
Header: ‘Discussion’ · subtitle ‘Descriptive patterns vs. one significant test’.
Three cards now cite t-test: A — empathy sig. lower than C (p = .017); B — trust descriptively highest, not sig.; C — empathy sig. higher than A.
Limitations box: n = 37, t-tests n = 35, 12 tests without Bonferroni correction.
Footer: 5 »
Slide 6 — Conclusion & Outlook
Recommendations: align disclosure to UX goal; don’t hide AI; B for trust / A for clarity — but note only empathy A vs. C was statistically significant.
Outlook: larger n, Bonferroni or ANOVA, real flows, Funke interactions.
Thank you · 6 »