0 cumulative citations
View corpus contextAI prompts reshape expert foresight: a small qualitative study finds that ChatGPT-generated inputs reorganize how healthcare experts sequence, frame and take positions on future scenarios — suggesting firms who pair domain experience with prompt skill may gain an edge.
Exploring Experts’ discourses in Corporate Foresight: Unveiling the temporal tension between past-present-future
openalex
quasi_experimental
low evidence
7/10 relevance
Summary only
summary available; pdf_status=paywall
DOI
Source
PDF
Exposure to ChatGPT-generated prompts changed how healthcare experts in Italy structured and reasoned about future scenarios, shifting temporal framing and evaluative stance rather than merely altering substantive content.
Citation observations
Cumulative provider counts captured on specific dates; providers are never combined.
0 cumulative citations
View corpus contextThe role of experts in foresight and corporate foresight has become increasingly debated, as empirical evidence suggests that expert predictions often underperform alternative approaches. This has shifted attention from what experts predict to how they construct and articulate futures. The objective of this study is to examine how experts discursively construct future-oriented narratives and how these processes are shaped by different types of informational input. Specifically, it advances four contributions: it conceptualizes expertise as a situated and practice-based activity, it reframes expertise as a performative and discursive accomplishment, it shows how temporal reasoning structures expert discourse, and it demonstrates how material inputs influence the way futures are constructed. Empirically, the study focuses on patient empowerment in healthcare, a domain undergoing significant technological and societal transformation. Drawing on a qualitative research design inspired by corporate foresight practices, we conducted semi-structured interviews with 21 Italian experts, divided into two groups, one of which was exposed to ChatGPT-generated foresight prompts. The findings show that experts construct futures through the dynamic mobilization of experience, contextual knowledge, and evaluative judgment. Moreover, AI-generated inputs do not simply affect the content of expert views, but shape how experts organize their reasoning and position themselves toward future developments. Overall, the study contributes to a more nuanced understanding of foresight as a practice and offers insights into how expert knowledge and material inputs jointly shape the construction of futures.
Summary
Main Finding
Experts construct futures through situated, practice-based discursive work that dynamically mobilizes experience, contextual knowledge, and evaluative judgment — and AI-generated informational inputs (ChatGPT prompts) do more than change content: they reorganize how experts reason about and position themselves toward future developments.
Key Points
- Expertise is framed as a situated, practice-based activity rather than a static possession.
- Expert foresight is performative and discursive: the way experts articulate futures is constitutive of those futures.
- Temporal reasoning (how experts sequence, pace, and horizon-set developments) is a core structuring device of expert discourse.
- Material inputs (here, AI-generated prompts) shape not only what experts say but how they organize reasoning and take stances toward future scenarios.
- Empirical setting: patient empowerment in healthcare undergoing technological and societal change; highlights domain-specific dynamics of foresight.
- AI-mediated inputs can shift experts’ reasoning patterns and evaluative frames, suggesting complementarities and interaction effects between human expertise and AI tools.
Data & Methods
- Design: Qualitative study inspired by corporate foresight practices.
- Participants: 21 Italian experts in patient empowerment / healthcare, split into two groups.
- Treatment: One group was exposed to ChatGPT-generated foresight prompts; the other was not.
- Method: Semi-structured interviews analyzing discursive construction of futures, temporal framing, and positioning.
- Analysis: Interpretive coding of narratives to identify how experiential, contextual, and evaluative resources are mobilized and how AI inputs alter discourse organization.
Implications for AI Economics
- Valuation of expertise: If AI inputs change the form (not just the content) of expert reasoning, market value of forecasts and advisory services may shift toward providers who combine domain experience with skillful use of AI prompts.
- Complementarity vs. substitution: Results suggest AI acts as a material complement that restructures expert cognition and articulation, implying models of labor demand should account for new complementarities (prompting skill, discursive framing) rather than pure substitution.
- Information design and market outcomes: AI-generated prompts can systematically reframe temporal expectations and risk perceptions, potentially altering investment, R&D prioritization, and pricing decisions in sectors where expert foresight influences markets (e.g., healthcare innovation).
- Measurement and incentives: Economic evaluations of forecasting accuracy should include measures of discursive structure and temporal framing, not only point predictions — incentives and contract design may need updating to reward useful framing and scenario construction.
- Policy and governance: Regulators and firms should be aware that AI tools can introduce framing effects and epistemic shifts; transparency about AI inputs, diversity of informational sources, and training in critical prompt use are warranted to mitigate biased or homogenized foresight.
- Research directions: Quantitative follow-ups (larger samples, accuracy metrics, cross-domain comparisons) and formal models incorporating performative/discursive effects of AI on expectations formation can inform forecasts of labor reallocation, pricing of advisory services, and systemic risk from synchronized expert narratives.
Limitations to consider when applying these implications: small, qualitative Italian sample focused on healthcare; effects may vary by domain, AI system, and institutional context.
Assessment
Paper Typequasi_experimental
Evidence Strengthlow — Small, qualitative sample with no clear randomization, heavy reliance on interpretive coding, and no objective behavioral or accuracy outcomes — the design supports suggestive, process-level claims about discourse but is weak for strong causal claims or generalization.
Methods Rigormedium — The study uses appropriate qualitative tools (semi-structured interviews, interpretive coding) and a between-group treatment contrast which is a useful exploratory design, but it lacks clear random assignment, pre-registration, inter-coder reliability reporting, triangulation with quantitative measures, and objective outcome metrics.
Sample21 Italian experts in patient empowerment/healthcare, purposively sampled and divided into two groups (one exposed to ChatGPT-generated foresight prompts, one not); data are semi-structured interview transcripts analyzed via interpretive coding. Recruitment, randomization, and participant characteristics beyond domain expertise are not detailed.
Themeshuman_ai_collab labor_markets org_design
IdentificationBetween-group qualitative comparison: 21 experts split into two groups, with one group exposed to ChatGPT-generated foresight prompts and the other not; causal inference rests on contrasting discursive patterns across treated vs control interviews and attributing observed differences to the AI-generated inputs.
GeneralizabilitySmall sample (N=21) limits statistical generalizability, Single country/cultural context (Italy) — cultural/institutional specifics may shape discourse, Single domain (patient empowerment/healthcare) — findings may not transfer to other sectors, Specific AI input (ChatGPT prompts) — other models/prompting regimes may produce different effects, Qualitative, interpretive outcomes (discursive structure) may not map directly to economic behaviors or market outcomes, Unclear sampling/randomization raises risk of selection bias
Claims (8)
| Claim | Direction | Outcome | Confidence & Evidence | Details |
|---|---|---|---|---|
| Experts construct futures through situated, practice-based discursive work that mobilizes experience, contextual knowledge, and evaluative judgment. Decision Quality | positive | Expert reasoning and construction of future scenarios |
Reading fidelity
high
Study strength
medium
|
n=21
|
| Expert foresight is performative and discursive, meaning that how experts articulate futures contributes to constituting those futures. Decision Quality | positive | Discursive construction and framing of future scenarios |
Reading fidelity
high
Study strength
medium
|
n=21
|
| Temporal reasoning—how experts sequence, pace, and set the horizon for developments—is a core structuring device of expert discourse. Decision Quality | positive | Temporal framing of future developments |
Reading fidelity
high
Study strength
medium
|
n=21
|
| Exposure to ChatGPT-generated foresight prompts changes not only the content of expert responses but also how experts organize their reasoning and position themselves toward future scenarios. Decision Quality | mixed | Organization of reasoning and evaluative positioning toward future scenarios |
Reading fidelity
high
Study strength
medium
|
n=21
|
| AI-generated prompts can shift experts’ reasoning patterns and evaluative frames, indicating complementarities and interaction effects between human expertise and AI tools. Task Allocation | positive | Interaction between AI inputs and expert reasoning |
Reading fidelity
high
Study strength
medium
|
n=21
|
| The study finds that AI-generated prompts function as material inputs that restructure expert cognition and articulation rather than acting solely as substitutes for expert judgment. Task Allocation | positive | Human-AI complementarity in expert reasoning |
Reading fidelity
high
Study strength
medium
|
n=21
|
| AI-generated prompts may systematically reframe temporal expectations and risk perceptions, with potential consequences for investment, R&D prioritization, and pricing decisions. Decision Quality | mixed | Investment, R&D-prioritization, and pricing decisions shaped by expert foresight |
Reading fidelity
medium
Study strength
speculative
|
n=21
|
| The empirical findings are limited in generalizability because they come from a small qualitative Italian sample focused on healthcare, and effects may vary by domain, AI system, and institutional context. Other | negative | External validity and generalizability of the findings |
Reading fidelity
high
Study strength
high
|
n=21
|