Responsible use of generative AI in Chinese vocational education: eye-tracking evidence on source verification and immediate near-transfer across three information-seeking conditions

May combine two claims.

Tool-supported performance must be distinguished from durable capability development: performance gains produced by GenAI do not necessarily constitute learning or skill acquisition.

“Tool-supported performance must also be distinguished from durable capability development: performance gains produced by GenAI do not necessarily constitute learning or skill acquisition (Fauth and González-Martínez, 2021; Melumad and Yun, 2025; Yan et al., 2025).”

• cites Fauth and González-Martínez, 2021 sourceNot checked — no full text of the source
• cites Melumad and Yun, 2025 sourceSupported
• cites Yan et al., 2025 sourceSupported

May combine two claims.

Technology acceptance, learning outcomes, and user attitudes have often been examined separately, leaving the process links among search mode, task trust, source verification, and learning transfer insufficiently explained.

“Second, technology acceptance, learning outcomes, and user attitudes have often been examined separately, leaving the process links among search mode, task trust, source verification, and learning transfer insufficiently explained (García-Alonso et al., 2024; Chen et al., 2025).”

• cites García-Alonso et al., 2024 sourceCould not verify

  The Literature Review (§2.3) discusses ChatGPT acceptance, attitudes, and perceived learning value, while the Results (§4) examine credibility and attitudes. Neither passage says these topics have often been studied separately or identifies an unexplained link among search mode, task trust, source verification, and learning transfer. Searches for those process links and the claimed research gap found no supporting passage in our copy. (LLM, conf=0.75)

  Experimental — accuracy not yet measured
• cites Chen et al., 2025 sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

Cognitive forcing functions can reduce overreliance on AI, although they also increase perceived effort.

“Cognitive forcing functions can reduce such overreliance, although they also increase perceived effort.”

• cites Buçinca et al. (2021) sourcePartly supported

  The Introduction reports that cognitive forcing functions significantly reduced overreliance compared with simple explainable AI approaches. But Section 4.2, “Subjective measures” (page unavailable), reports no increase in rated mental demand: the cognitive-forcing and simple-explanation conditions had nearly identical scores, both below the no-AI condition. Participants did rate the cognitive-forcing systems as more complex, which is distinct from perceived effort. (LLM, conf=0.90)

  Experimental — accuracy not yet measured

May combine two claims.

The speed, fluency, and integrated presentation of GenAI output can reduce operational friction while increasing heuristic trust and reducing spontaneous verification.

“From this perspective, the speed, fluency, and integrated presentation of GenAI output can reduce operational friction while increasing heuristic trust and reducing spontaneous verification (Lee and See, 2004; Vasconcelos et al., 2023).”

• cites Lee and See, 2004 sourceNot checked — no full text of the source
• cites Vasconcelos et al., 2023 sourcePartly supported

  The Introduction says people may rely on an AI prediction without checking it when verification is cognitively costly, supporting the claim’s link between effort and reduced verification. But the study used a simulated AI in a maze task, not GenAI output; it does not establish effects of speed, fluency, or integrated presentation on heuristic trust. The Limitations and Future Work section says measuring and modulating trust was outside its scope. (LLM, conf=0.87)

  Experimental — accuracy not yet measured

May combine two claims.

The polished appearance of AI-generated content may make it more difficult for learners to distinguish readability from reliability, encouraging them to equate expressive quality with evidential quality and reducing their motivation to verify.

“Its polished appearance may make it more difficult for learners to distinguish readability from reliability (Kasneci et al., 2023; Lee et al., 2025), encouraging them to equate expressive quality with evidential quality and reducing their motivation to verify (Anderl et al., 2024; Martín-Moncunill and Alonso Martínez, 2025).”

• cites Kasneci et al., 2023 sourceSupported
• cites Lee et al., 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Anderl et al., 2024 sourcePartly supported

  Split into 2 atoms — 0 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 1 unverifiable. Expand the claim to see per-atom verdicts.

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Buçinca et al. (2021) showed that users may rely on a generalized heuristic trust in AI rather than analytically evaluating each recommendation.

“In AI-assisted decision making, Buçinca et al. (2021) further showed that users may rely on a generalized heuristic trust in AI rather than analytically evaluating each recommendation.”

• cites Buçinca et al. (2021) sourceSupported

May combine two claims.

Bastani et al. (2025) showed that unconstrained GenAI support could improve performance while the tool was available yet undermine subsequent independent performance, whereas learning safeguards mitigated this effect.

“Bastani et al. (2025) showed that unconstrained GenAI support could improve performance while the tool was available yet undermine subsequent independent performance, whereas learning safeguards mitigated this effect.”

• cites Bastani et al. (2025) sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Qian et al. (2026) suggests that the effects of GenAI on cognitive load depend on scaffolding, intensity of use, prior knowledge, and task design.

“Recent review evidence similarly suggests that the effects of GenAI on cognitive load depend on scaffolding, intensity of use, prior knowledge, and task design (Qian et al., 2026).”

• cites Qian et al., 2026 sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing).

Internal consistency of the questionnaire scales was assessed before the primary analyses using Cronbach's α.

“Internal consistency was assessed before the primary analyses using Cronbach’s α (Taber, 2018).”

• cites Taber, 2018 sourceSupported

May combine two claims.

The sample provided adequate power to detect between-group effects of moderate magnitude or larger, although its capacity to identify small effects remained limited.

“The sample therefore provided adequate power to detect between group effects of moderate magnitude or larger, although its capacity to identify small effects remained limited (Kang, 2021).”

• cites Kang, 2021 sourcePartly supported

  In “Examples 1. Two-sample or independent t-test” (page not available), Kang calculates that 64 participants per group provide 80% power for a medium effect (d = 0.5). This supports the general relationship between sample size, effect size, and power, but the paper does not assess the citing study’s sample or establish that it had adequate power for moderate effects and limited power for small ones. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

Permuted-block randomization followed established allocation principles for balancing treatment groups.

“Permuted-block randomization followed established allocation principles for balancing treatment groups (Lim and In, 2019).”

• cites Lim and In, 2019 sourceSupported

Classified as probably not a substantial claim.

The AOI classification enabled attention to equivalent information functions to be compared despite differences in page layout and interface structure.

“This classification enabled attention to equivalent information functions to be compared despite differences in page layout and interface structure (Hooge et al., 2026).”

• cites Hooge et al., 2026 sourcePartly supported

  In ‘Static AOIs > Deciding on the AOI’ (page not provided), the paper describes defining AOIs by information category and discusses comparing the same objects across different images or displays. It does not establish that the citing authors’ classification enabled comparison of equivalent information functions despite differences in page layout and interface structure. (LLM, conf=0.79)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim. May combine two claims.

Stadler et al. (2024) reported that LLMs can reduce the burden of information seeking even when lower cognitive effort is accompanied by shallower reasoning.

“This interpretation is consistent with Stadler et al. (2024), who reported that LLMs can reduce the burden of information seeking even when lower cognitive effort is accompanied by shallower reasoning.”

• cites Stadler et al. (2024) sourceSupported

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Buçinca et al. (2021) found that cognitive forcing functions reduced uncritical reliance on AI advice while increasing perceived effort.

“Buçinca et al. (2021) similarly found that cognitive forcing functions reduced uncritical reliance on AI advice while increasing perceived effort.”

• cites Buçinca et al. (2021) sourceSupported

May combine two claims.

Hybrid search achieved the highest immediate near-transfer score (86.99) and outperformed both other conditions on near-transfer performance.

“and immediate near-transfer score (86.99), and outperformed both other conditions on near-transfer performance”

May combine two claims.

The hybrid-search condition showed broader visual coverage across the task-material area while retaining a distinct concentration of fixations within the task-requirement area.

“By contrast, the hybrid-search condition showed broader visual coverage across the task-material area while retaining a distinct concentration of fixations within the task-requirement area.”

May combine two claims.

Hybrid search significantly outperformed GenAI-only search on both information-integration and near-transfer performance and also outperformed conventional web search on near-transfer performance.

“Hybrid search significantly outperformed GenAI-only search on both information-integration and near-transfer performance and also outperformed conventional web search on near-transfer performance.”

May combine two claims.

Hybrid search had intermediate operational costs and was associated with greater source attention, more answer revision, and better immediate near-transfer performance.

“Hybrid search had intermediate operational costs and was associated with greater source attention, more answer revision, and better immediate near-transfer performance.”

• Also stated in: § Conclusion · plain

  “Hybrid search had intermediate operational costs and produced greater source-related attention, more answer revision, and better condition-supported immediate near-transfer performance.”

May combine two claims.

In hybrid search, trust was lower but verification and revision were substantially more frequent.

“In hybrid search, trust was lower but verification and revision were substantially more frequent.”

May combine two claims.

GenAI-only participants had the shortest combined active completion time across the two timed tasks but showed less cross-verification and answer revision and more direct copying, whereas hybrid search maintained relatively high verification and revision with less time than conventional web search.

“GenAI-only participants had the shortest combined active completion time across the two timed tasks but showed less cross-verification and answer revision and more direct copying, whereas hybrid search maintained relatively high verification and revision with less time than conventional web search.”

May combine two claims.

In the GenAI-only condition, the 25.0% verification rate represented requests for a second AI-provided source concerning the same claim; independent external verification was possible only in the web-enabled conditions.

“In the GenAI-only condition, the 25.0% verification rate represented requests for a second AI-provided source concerning the same claim; independent external verification was possible only in the web-enabled conditions.”

May combine two claims.

Conventional web search was the most demanding, but it preserved more extensive multi-source browsing.

“Conventional web search was the most demanding, but it preserved more extensive multi-source browsing.”

• Also stated in: § Conclusion · plain

  “Conventional web search required the most time and produced the greatest perceived load, while preserving more multi-source browsing and evidence comparison.”

May combine two claims.

Eye-tracking and behavioral results showed less source-evidence AOI dwell, fewer source revisits and cross-region transitions, and lower rates of cross-verification and answer revision in the GenAI-only condition.

“Eye-tracking and behavioral results showed less source-evidence AOI dwell, fewer source revisits and cross-region transitions, and lower rates of cross-verification and answer revision in the GenAI-only condition.”

May combine two claims.

In the GenAI-only condition, although task trust was highest, source-evidence attention, cross-verification, and answer revision were lower and direct copying was higher than in the hybrid-search condition.

“Although task trust was highest, source-evidence attention, cross-verification, and answer revision were lower and direct copying was higher than in the hybrid-search condition (Tables 3 and 4).”

• AssumptionThe authors state this as an assumption: The absence of a source evidence AOI in a page state reflects an interface characteristic rather than the participant ignoring available source information.

May combine two claims.

Source-evidence dwell, related revisits, and transitions were lowest in the GenAI-only condition and highest in the hybrid-search condition.

“The clearest pattern concerned attention to source evidence: source-evidence dwell and related revisits and transitions were lowest in the GenAI-only condition and highest in the hybrid-search condition.”

• AssumptionThe authors state this as an assumption: The absence of a source evidence AOI in a page state reflects an interface characteristic rather than the participant ignoring available source information.

May combine two claims.

Task trust was highest in the GenAI-only condition, lowest in the conventional web-search condition, and intermediate in the hybrid-search condition.

“It was highest in the GenAI-only condition, which had the lowest near-transfer score, lowest in the conventional web-search condition, and intermediate in the hybrid-search condition.”

May combine two claims.

Overall, the representative heatmaps suggest more localized task-page attention in the GenAI-only and conventional web-search conditions and broader visual coverage in the hybrid-search condition.

“Overall, the representative heatmaps suggest more localized task-page attention in the GenAI-only and conventional web-search conditions and broader visual coverage in the hybrid-search condition (Figure 6).”

May combine two claims.

GenAI-only search was the fastest and produced the lowest overall perceived cognitive load.

“GenAI-only search was the fastest and produced the lowest overall perceived cognitive load.”

• Also stated in: § Conclusion · plain

  “GenAI-only search was the fastest and produced the lowest overall perceived cognitive load, but it was accompanied by less attention to source evidence, fewer cross-source verification attempts, and fewer answer revisions.”

May combine two claims.

GenAI-only participants concentrated more attention within the information-acquisition region, whereas hybrid-search participants shifted more often between content, source evidence, and response construction.

“GenAI-only participants concentrated more attention within the information-acquisition region, whereas hybrid-search participants shifted more often between content, source evidence, and response construction.”

• AssumptionThe authors state this as an assumption: The absence of a source evidence AOI in a page state reflects an interface characteristic rather than the participant ignoring available source information.

Probably not a substantial claim: may not pass S3 (load-bearing).

These descriptive visualizations were consistent in direction with the group-level eye-tracking and behavioral results.

“These descriptive visualizations were consistent in direction with the group-level eye-tracking and behavioral results, but they were used only to illustrate possible information-processing pathways, as shown in Figure 7.”

May combine two claims.

Task trust did not align uniformly with performance and showed no stable positive relationship with immediate near-transfer performance.

“Task trust likewise did not align uniformly with performance.”

• Also stated in: § Conclusion · plain

  “Task trust showed no stable positive relationship with immediate near-transfer performance.”

May combine two claims.

Perceived verification effort was highest under hybrid search but did not differ significantly from conventional web search.

“Perceived verification effort was highest under hybrid search but did not differ significantly from conventional web search.”

May combine two claims.

The conventional web-search condition showed more switching between search results and webpage content, whereas the hybrid-search condition showed more frequent cross-region revisits, particularly among the generated response, source information, and webpage content.

“The conventional web-search condition showed more switching between search results and webpage content, whereas the hybrid-search condition showed more frequent cross-region revisits, particularly among the generated response, source information, and webpage content.”

May combine two claims.

After adjustment for search condition and prior tool experience, hybrid search (β = 0.24, 95% CI [0.04, 0.44], p = 0.019), source-evidence AOI dwell proportion (β = 0.28, 95% CI [0.08, 0.48], p = 0.006), and cross-verification attempts (β = 0.31, 95% CI [0.12, 0.50], p = 0.002) were positively associated with near-transfer performance.

“After adjustment for search condition and prior tool experience, hybrid search (β = 0.24, 95% CI [0.04, 0.44], p = 0.019), source-evidence AOI dwell proportion (β = 0.28, 95% CI [0.08, 0.48], p = 0.006), and cross-verification attempts (β = 0.31, 95% CI [0.12, 0.50], p = 0.002) were positively associated with near-transfer performance”

• AssumptionThe authors state this as an assumption: The absence of a source evidence AOI in a page state reflects an interface characteristic rather than the participant ignoring available source information.

May combine two claims.

Omnibus tests were significant for all six outcomes: task trust, overall perceived cognitive load, perceived verification effort, metacognitive monitoring, information-integration performance, and immediate near-transfer performance.

“Omnibus tests were significant for all six outcomes; complete descriptive statistics, effect sizes, and adjusted pairwise comparisons are reported in Table 4.”

May combine two claims.

In the conventional web-search condition, fixations were concentrated mainly in the upper portion of the task-material area, whereas attention to the task-requirement area was relatively limited and dispersed.

“In the conventional web-search condition, fixations were concentrated mainly in the upper portion of the task-material area, whereas attention to the task-requirement area was relatively limited and dispersed.”

May combine two claims.

GenAI-only search produced the highest task trust and the lowest overall perceived cognitive load, whereas hybrid search produced the highest metacognitive monitoring and the strongest immediate near-transfer performance.

“GenAI-only search produced the highest task trust and the lowest overall perceived cognitive load, whereas hybrid search produced the highest metacognitive monitoring and the strongest immediate near-transfer performance.”

• AssumptionThe authors state this as an assumption: Overall perceived cognitive load measured once after both tasks reflects the general mental effort and task-processing burden experienced during the assigned search-and-response workflow.

May combine two claims.

The exploratory temporal plot was consistent with the pattern that GenAI-only participants moved quickly to the AI interface, conventional web-search participants progressed toward external webpages, and hybrid-search participants alternated more often among the AI interface, search results, and source webpages.

“The exploratory temporal plot was consistent with this pattern: GenAI-only participants moved quickly to the AI interface, conventional web-search participants progressed toward external webpages, and hybrid-search participants alternated more often among the AI interface, search results, and source webpages (Figure 5).”

May combine two claims.

Source clicks no longer explained unique variance after more substantive indicators were included, whereas attention to source evidence and cross-verification attempts remained associated with near-transfer performance.

“Source clicks no longer explained unique variance after more substantive indicators were included, whereas attention to source evidence and cross-verification attempts remained associated with near-transfer performance.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Research on technology adoption indicates that effort expectancy, social influence, and facilitating conditions shape students' intentions to use GenAI.

“Research on technology adoption likewise indicates that effort expectancy, social influence, and facilitating conditions shape students’ intentions to use GenAI (BinJwair, 2025).”

• cites BinJwair, 2025 sourceNot checked — no full text of the source

Classified as probably not a substantial claim. May combine two claims.

Dual-process accounts propose that fast, automatic Type 1 processing typically generates a default response, whereas working-memory-dependent Type 2 processing becomes more fully engaged when more deliberate judgment is required.

“Dual-process accounts propose that fast, automatic Type 1 processing typically generates a default response, whereas working-memory-dependent Type 2 processing becomes more fully engaged when more deliberate judgment is required (Evans and Stanovich, 2013).”

• cites Evans and Stanovich, 2013 sourceNot checked — no full text of the source

May combine two claims.

GenAI may compress the multi-source comparison and evidence-verification processes through which learning occurs, increasing the risks of cognitive dependence, acceptance of inaccurate information, and weakened critical judgment.

“At the same time, it may compress the multi-source comparison and evidence-verification processes through which learning occurs, increasing the risks of cognitive dependence, acceptance of inaccurate information, and weakened critical judgment (Lee et al., 2025).”

• cites Lee et al., 2025Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Participants were stratified by year of study, disciplinary cluster, and previous AI use experience.

“participants were stratified by year of study, disciplinary cluster, and previous AI use experience”

Classified as probably not a substantial claim.

Within each stratum, participants were assigned to the three experimental conditions using computer-generated permuted-block randomization.

“Within each stratum, participants were assigned to the three experimental conditions using computer-generated permuted-block randomization.”

Classified as probably not a substantial claim.

Multimodal learning analytics can strengthen inference by triangulating gaze with recorded action and performance evidence.

“Multimodal learning analytics can strengthen this inference by triangulating gaze with recorded action and performance evidence (Järvelä et al., 2021; Mohammadi et al., 2025).”

• cites Järvelä et al., 2021 sourceNot checked — no full text of the source
• cites Mohammadi et al., 2025 sourceNot checked — no full text of the source

May combine two claims.

GenAI-only search reduced completion time and overall perceived cognitive load.

“GenAI-only search reduced completion time and overall perceived cognitive load”

May combine two claims.

Hybrid search was associated with more source checking and better immediate near-transfer performance under the assigned condition.

“whereas hybrid search was associated with more source checking and better immediate near-transfer performance under the assigned condition”

Classified as probably not a substantial claim.

Attention concentrated on answer or content areas, with source cues largely neglected, may indicate limited source awareness and shallower evidential judgment.

“Conversely, attention concentrated on answer or content areas, with source cues largely neglected, may indicate limited source awareness and shallower evidential judgment. This interpretation follows prior work on online source evaluation and gaze-based indicators of credibility assessment (Gottschling and Kammerer, 2021; Guan and Lin, 2025; Tsai et al., 2022).”

• cites Gottschling and Kammerer, 2021 sourceNot checked — no full text of the source
• cites Guan and Lin, 2025 sourceNot checked — no full text of the source
• cites Tsai et al., 2022 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Hybrid search may preserve advantages of both GenAI and conventional search by allowing learners to move flexibly among AI generation, keyword expansion, web search, source verification, and answer revision as the task requires.

“Hybrid search may preserve advantages of both modes by allowing learners to move flexibly among AI generation, keyword expansion, web search, source verification, and answer revision as the task requires. This combination may offer a workable balance between operational cost and evidential judgment (Boetje et al., 2026; Yen et al., 2024).”

• cites Boetje et al., 2026 sourceNot checked — no full text of the source
• cites Yen et al., 2024Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim. May combine two claims.

Lee et al. (2025) found that greater confidence in GenAI among knowledge workers was associated with less critical thinking and that, in GenAI-supported work, critical thinking shifted toward information verification, response integration, and task management.

“Lee et al. (2025) found that greater confidence in GenAI among knowledge workers was associated with less critical thinking and that, in GenAI-supported work, critical thinking shifted toward information verification, response integration, and task management.”

• cites Lee et al. (2025)Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

Classified as probably not a substantial claim.

Source cues alone do not demonstrate that verification has occurred.

“Source cues alone, however, do not demonstrate that verification has occurred.”

May combine two claims.

The sensitivity analysis within the two web-enabled conditions reduced this confounding but could not decompose the complete workflow effect into pure tool and motivational components.

“The sensitivity analysis within the two web-enabled conditions reduced this confounding but could not decompose the complete workflow effect into pure tool and motivational components.”

Classified as probably not a substantial claim.

For cross-condition analyses, functionally comparable regions were aggregated into four higher-level categories: task comprehension, information acquisition, source evidence, and answer construction.

“For cross condition analyses, functionally comparable regions were aggregated into four higher level categories: the task comprehension AOI, information acquisition AOI, source evidence AOI, and answer construction AOI.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Heatmaps and scanpaths were used only to support descriptive interpretation of visual-search strategies and were not used as the primary basis for significance testing.

“Heatmaps and scanpaths were used only to support descriptive interpretation of visual-search strategies; they were not used as the primary basis for significance testing.”

Classified as probably not a substantial claim.

Before each session, the browser cache, search history, and login status were cleared to minimize the influence of personalized recommendations on the search results.

“Before each session, the browser cache, search history, and login status were cleared to minimize the influence of personalized recommendations on the search results.”

Classified as probably not a substantial claim.

Eye-tracking measures were interpreted jointly with behavioral logs, cross-verification, answer revision, and task performance.

“They were therefore interpreted jointly with behavioral logs, cross-verification, answer revision, and task performance.”

Classified as probably not a substantial claim. May combine two claims.

All three conditions completed the same task under the same time limit.

“All three conditions completed the same task under the same time limit.”

Classified as probably not a substantial claim. May combine two claims.

Because source-access measures were shaped by condition-specific affordances, the zero-order correlations are descriptive; hierarchical regression provides the primary adjusted estimates.

“Because source-access measures were shaped by condition-specific affordances, these correlations are descriptive; the hierarchical regression provides the primary adjusted estimates.”

Classified as probably not a substantial claim. May combine two claims.

Source-evidence AOI dwell proportion, revisits, and transitions captured attention to evidence and movement among functional regions, while cross-verification and answer revision indicated whether judgments changed in response to evidence.

“Source-evidence AOI dwell proportion, revisits, and transitions captured attention to evidence and movement among functional regions, while cross-verification and answer revision indicated whether judgments changed in response to evidence.”

• AssumptionRests on an assumption: The eye-tracking and behavioral measures used (source-evidence AOI dwell, revisits, transitions, cross-verification attempts, answer revision) validly index the cognitive processes of source verification and evidence evaluation.

May combine two claims.

The experimental task simplified extended learning and authentic workplace demands and included an explicit verification requirement that may have elevated verification behavior above everyday GenAI use.

“Although the experimental task approximated a vocational context, it simplified extended learning and authentic workplace demands and included an explicit verification requirement that may have elevated verification behavior above everyday GenAI use.”

Classified as probably not a substantial claim.

AOIs were coded according to the content visible in each recorded page state rather than treated as permanently fixed screen coordinates.

“AOIs were therefore coded according to the content visible in each recorded page state rather than treated as permanently fixed screen coordinates.”

Classified as probably not a substantial claim. May combine two claims.

Behavioral logs were independently coded by two coders, and task responses were independently evaluated by two raters; intercoder and interrater reliability were calculated separately.

“Behavioral logs were independently coded by two coders, and task responses were independently evaluated by two raters. Intercoder and interrater reliability were calculated separately.”

Classified as probably not a substantial claim. May combine two claims.

Task materials, time limits, response requirements, and verification objectives were identical across the three conditions.

“Task materials, time limits, response requirements, and verification objectives were identical across the three conditions.”

Classified as probably not a substantial claim.

The conditions differed only in the information-access tools available and the resulting affordances for accessing sources.

“The conditions differed only in the information-access tools available and the resulting affordances for accessing sources.”

Classified as probably not a substantial claim.

These measures cannot independently establish critical evaluation or deep understanding.

“These measures, however, cannot independently establish critical evaluation or deep understanding.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Prior AI-use experience and web-search experience were retained as covariates in the regression models to account for individual differences in familiarity with the two information-seeking tools.

“Prior AI-use experience and web-search experience were nevertheless retained as covariates in the regression models to account for individual differences in familiarity with the two information-seeking tools.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

In the recruited sample, exclusions comprised insufficient valid gaze data below 75% (n = 4), eye-tracker calibration failure related to severe myopia or other eye difficulties (n = 1), and a protocol violation involving unauthorized AI use (n = 1).

“In the recruited sample, exclusions comprised insufficient valid gaze data below 75% (n = 4), eye-tracker calibration failure related to severe myopia or other eye difficulties (n = 1), and a protocol violation involving unauthorized AI use (n = 1).”

May combine two claims.

Because the design did not establish the temporal ordering required for causal mediation, no causal mediation model with task trust as a mediator was estimated; zero-order correlations and multivariable regression were used to characterize statistical associations.

“Because the design did not establish the temporal ordering required for causal mediation, we did not estimate a causal mediation model with task trust as a mediator. Instead, zero-order correlations and multivariable regression were used to characterize statistical associations among overall perceived cognitive load, task trust, verification behavior, and immediate near-transfer performance.”

Classified as probably not a substantial claim. May combine two claims.

The proportions reported across all three conditions describe cross-source verification attempts in a broad, affordance-specific sense; their implementations were not equivalent to independent external verification.

“The proportions reported across all three conditions therefore describe cross-source verification attempts in a broad, affordance-specific sense; their implementations were not equivalent to independent external verification.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Each participant completed an eye-tracker calibration before the formal experiment.

“Each participant completed an eye tracker calibration before the formal experiment.”

Classified as probably not a substantial claim. May combine two claims.

In the GenAI-only condition, the 25.0% cross-verification-attempt rate specifically denotes requests to GenAI for a second source for the same claim; independent external comparison was possible only in the web-enabled conditions.

“In the GenAI-only condition, the 25.0% cross-verification-attempt rate specifically denotes requests to GenAI for a second source for the same claim; independent external comparison was possible only in the web-enabled conditions.”

May combine two claims.

Hybrid search's advantage lay primarily in the verification process and immediate performance, not operational efficiency or overall trust.

“Its advantage lay primarily in the verification process and immediate performance, not operational efficiency or overall trust.”

Probably not a substantial claim: may not pass S3 (load-bearing).

No participant was excluded for withdrawal.

“No participant was excluded for withdrawal.”

Classified as probably not a substantial claim.

In the GenAI-only condition, asking GenAI to provide a second source for the same factual claim was coded as a cross-source verification attempt.

“Because participants in the GenAI-only condition could not independently access external webpages, asking GenAI to provide a second source for the same factual claim was coded as a cross-source verification attempt.”

Classified as probably not a substantial claim. May combine two claims.

The task materials, time limits, response requirements, and laboratory environment were held constant across all three conditions; the conditions differed only in the forms of information-seeking support available to participants.

“The task materials, time limits, response requirements, and laboratory environment were held constant across all three conditions; the conditions differed only in the forms of information-seeking support available to participants.”

Probably not a substantial claim: may not pass S3 (load-bearing).

Additional participants were recruited to account for calibration failure, withdrawal, protocol violations, and valid gaze data rates below 75%.

“Additional participants were recruited to account for calibration failure, withdrawal, protocol violations, and valid gaze data rates below 75%.”

Classified as probably not a substantial claim.

The study was designed to retain at least 30 valid participants in each condition.

“the study was designed to retain at least 30 valid participants in each condition”

Classified as probably not a substantial claim.

Because the conditions did not provide equivalent access to independent external sources, regression findings involving cross-verification were interpreted in light of each condition's affordances.

“Because the conditions did not provide equivalent access to independent external sources, regression findings involving cross-verification were interpreted in light of each condition’s affordances.”

May combine two claims.

The formal task explicitly required participants in all conditions to locate, evaluate, and verify information, while source-access channels differed by condition.

“Moreover, the formal task explicitly required participants in all conditions to locate, evaluate, and verify information, while source-access channels differed by condition.”

Classified as probably not a substantial claim.

To reduce the influence of variation in the number of pages visited and total browsing time, AOI measures were normalized relative to each participant's total valid fixation duration or valid viewing time.

“To reduce the influence of variation in the number of pages visited and total browsing time, AOI measures were normalized relative to each participant’s total valid fixation duration or valid viewing time, as appropriate.”

Classified as probably not a substantial claim.

In the conventional web-search and hybrid-search conditions, independent cross-source verification was coded when a participant accessed and compared at least two mutually independent external sources.

“In the conventional web-search and hybrid-search conditions, independent cross-source verification was coded when a participant actually accessed and compared at least two mutually independent external sources.”

Classified as probably not a substantial claim. May combine two claims.

Eye-tracking measures by themselves indicate only visual attention and viewing behavior; they cannot establish that participants understood the evidence or engaged in critical evaluation.

“By themselves, however, they indicate only visual attention and viewing behavior; they cannot establish that participants understood the evidence or engaged in critical evaluation.”

Classified as probably not a substantial claim. May combine two claims.

Two raters independently evaluated all responses using a standardized scoring rubric and completed calibration training before formal scoring.

“Two raters independently evaluated all responses using a standardized scoring rubric and completed calibration training before formal scoring.”

May combine two claims.

The higher verification rate in the hybrid-search condition reflects the combined influence of the verification requirement and tool affordances and should not be equated with a spontaneous verification tendency in naturalistic use.

“Accordingly, the higher verification rate in the hybrid-search condition reflects the combined influence of the verification requirement and tool affordances and should not be equated with a spontaneous verification tendency in naturalistic use.”

• Also stated in: § Conclusion · plain

  “Because source-access affordances differed across conditions, the observed differences reflect the combined effects of complete workflows and tool availability and cannot be attributed solely to verification motivation.”

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

These findings frame responsible use as a process that can be observed and shaped through design: AI can support generation and organization, while learners retain responsibility for verification, revision, and final judgment.

“These findings frame responsible use as a process that can be observed and shaped through design: AI can support generation and organization, while learners retain responsibility for verification, revision, and final judgment.”

May combine two claims.

The study provides evidence about verification behavior and trust calibration, not a direct test of learners' ability to identify low-quality generated content.

“It therefore provides evidence about verification behavior and trust calibration, not a direct test of learners’ ability to identify low-quality generated content.”

Classified as probably not a substantial claim.

Multicollinearity among predictors was assessed using tolerance and the variance inflation factor to confirm that source clicks, cross-verification, and other predictors could be included simultaneously in the final model.

“Multicollinearity among predictors was assessed using tolerance and the variance inflation factor (VIF) to confirm that source clicks, cross-verification, and the other predictors could be included simultaneously in the final model.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Inter-rater reliability was calculated after scoring; when reliability met the predefined criterion, the mean of the two ratings was used as the final score, and cases with substantial disagreement were reviewed by a third rater.

“Inter rater reliability was calculated after scoring. When reliability met the predefined criterion, the mean of the two ratings was used as the final score. Cases involving substantial disagreement were reviewed by a third rater.”

Probably not a substantial claim: may not pass S2 (checkable), S4 (based).

If citations serve primarily as signals of credibility, learners may infer that cited content is necessarily reliable.

“If citations serve primarily as signals of credibility, learners may infer that cited content is necessarily reliable.”

Classified as probably not a substantial claim.

The analyses concern observable verification behavior rather than a separately measured self-report construct of need for source verification.

“The analyses concern observable verification behavior rather than a separately measured self-report construct of need for source verification.”

Classified as probably not a substantial claim.

To avoid conflating different levels of behavior under the general label of verification, the study distinguished source access, a verification attempt, and independent cross-source verification.

“To avoid conflating different levels of behavior under the general label of “verification,” we distinguished source access, a verification attempt, and independent cross-source verification.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Of the six excluded participants, four had valid gaze data below the prespecified 75% threshold, one could not complete eye-tracker calibration, and one violated the assigned information-seeking protocol by using AI when it was not permitted.

“Of the six excluded participants, four had valid gaze data below the prespecified 75% threshold, one could not complete eye-tracker calibration because severe myopia or other eye-related difficulties prevented reliable calibration, and one violated the assigned information-seeking protocol by using AI when it was not permitted.”

Classified as probably not a substantial claim.

Gaze data across pages were aggregated by functionally equivalent AOI category rather than compared on the basis of individual page layouts or fixed screen positions.

“Gaze data across pages were aggregated by functionally equivalent AOI category rather than compared on the basis of individual page layouts or fixed screen positions.”

May combine two claims.

GenAI reduced the operational effort associated with iterative searching or query reformulation, result screening, multi-page navigation, and information organization, but may also have compressed source tracing, evidence comparison, and judgment of applicability.

“GenAI reduced the operational effort associated with iterative searching or query reformulation, result screening, multi-page navigation, and information organization, but may also have compressed source tracing, evidence comparison, and judgment of applicability.”

May combine two claims.

Among retained participants, the mean proportion of valid gaze data was 88.64% (SD = 5.73%), exceeding the predefined threshold of 75%.

“Among the retained participants, the mean proportion of valid gaze data was 88.64%, SD = 5.73%, exceeding the predefined threshold of 75%.”

Classified as probably not a substantial claim. May combine two claims.

Participants were excluded if calibration failed, substantial gaze drift was observed, or the proportion of valid gaze data fell below the predefined threshold of 75%.

“Participants were excluded if calibration failed, substantial gaze drift was observed, or the proportion of valid gaze data fell below the predefined threshold of 75%.”

May combine two claims.

External-source affordances differed across the three conditions, meaning that differences in cross-verification reflect both behavioral choice and tool availability.

“Second, external-source affordances differed across the three conditions, meaning that differences in cross-verification reflect both behavioral choice and tool availability.”

May combine two claims.

These results do not establish that hybrid search incurred no additional cognitive cost, nor can overall perceived load be treated as a direct indicator of deep processing.

“These results do not establish that hybrid search incurred no additional cognitive cost, nor can overall perceived load be treated as a direct indicator of deep processing.”

Classified as probably not a substantial claim. May combine two claims.

In RQ2, operational efficiency was defined exclusively by combined active completion time across the two timed tasks and overall perceived cognitive load; verification, task performance, and task trust were represented separately.

“To avoid conflating efficiency with verification quality or learning outcomes, operational efficiency was defined exclusively by combined active completion time across the two timed tasks and overall perceived cognitive load. Verification was represented separately by source-evidence AOI dwell, revisits, cross-source verification attempts, and answer revision; task performance was represented by”

• AssumptionThe authors state this as an assumption: Overall perceived cognitive load measured once after both tasks reflects the general mental effort and task-processing burden experienced during the assigned search-and-response workflow.

May combine two claims.

Baseline balance tests revealed no significant differences among the three conditions in gender, age, prior AI use experience, web search experience, AI acceptance, or baseline task trust, all p > 0.05, supporting broad comparability following stratified randomization.

“Baseline balance tests revealed no significant differences among the three conditions in gender, age, prior AI use experience, web search experience, AI acceptance, or baseline task trust, all p > 0.05, supporting the broad comparability following stratified randomization.”

Classified as probably not a substantial claim. May combine two claims.

The same test account was used throughout the experiment, with memory disabled or all previous conversations cleared before each session.

“The same test account was used throughout the experiment, with memory disabled or all previous conversations cleared before each session.”

May combine two claims.

The study did not directly manipulate AI-generated "slop," model hallucination rates, or source traceability.

“The study also did not directly manipulate AI-generated “slop,” model hallucination rates, or source traceability.”

Classified as probably not a substantial claim.

The AOI normalization procedure enabled attention to functionally equivalent information regions to be aggregated across page transitions, scrolling events, and changing interface layouts.

“This procedure enabled attention to functionally equivalent information regions to be aggregated across page transitions, scrolling events, and changing interface layouts.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

This division of labor is consistent with human-centered AI principles emphasizing human control, system reliability, and retained accountability.

“This division of labor is consistent with human-centered AI principles emphasizing human control, system reliability, and retained accountability (Amershi et al., 2019; Shneiderman, 2020).”

• cites Amershi et al., 2019Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

• cites Shneiderman, 2020 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Four experts in vocational education and educational technology reviewed the candidate tasks, and a pilot test was conducted with 15 students.

“To reduce potential confounding due to differences in task-material type, four experts in vocational education and educational technology reviewed the candidates, and a pilot test was conducted with 15 students.”

Classified as probably not a substantial claim.

The absence of a stable positive association between task trust and immediate near-transfer performance suggests that responsible human-AI collaboration should prioritize trust calibrated to the available evidence rather than simply increasing trust in AI.

“The absence of a stable positive association between task trust and immediate near-transfer performance likewise suggests that responsible human-AI collaboration should prioritize trust calibrated to the available evidence rather than simply increasing trust in AI.”

Classified as probably not a substantial claim. May combine two claims.

The low, moderate, and high AI-experience categories served only as pre-randomization strata and did not constitute additional treatment conditions.

“The low, moderate, and high AI-experience categories likewise served only as pre-randomization strata and did not constitute additional treatment conditions.”

May combine two claims.

The value of hybrid search lay in enabling learners to move through a sequence of generating, tracing, comparing, and revising information and to repeat that sequence in a structurally similar task.

“The value of hybrid search lay in enabling learners to move through a sequence of generating, tracing, comparing, and revising information and to repeat that sequence in a structurally similar task.”

May combine two claims.

The rapid and coherent output of GenAI may encourage heuristic acceptance with relatively little cognitive effort, whereas source access, explicit verification requirements, and comparison across tools can introduce limited analytic processing at critical evidence checkpoints.

“The rapid and coherent output of GenAI may encourage heuristic acceptance with relatively little cognitive effort, whereas source access, explicit verification requirements, and comparison across tools can introduce limited analytic processing at critical evidence checkpoints.”

May combine two claims.

Lower overall perceived cognitive load does not automatically indicate a better learning process, just as greater load does not necessarily indicate deeper processing.

“Lower overall perceived cognitive load does not automatically indicate a better learning process, just as greater load does not necessarily indicate deeper processing.”

Classified as probably not a substantial claim.

Because GenAI interaction pages, search engine results pages, and external webpages differ substantially in visual structure, AOIs were defined according to functional equivalence rather than fixed spatial coordinates.

“Because GenAI interaction pages, search engine results pages, and external webpages differ substantially in their visual structure, AOIs were defined according to functional equivalence rather than fixed spatial coordinates.”

May combine two claims.

The final sample comprised 96 participants, with 32 assigned to each of the GenAI-only condition, conventional web search, and hybrid search conditions.

“The final sample comprised 96 participants, with 32 assigned to each of the GenAI-only condition, conventional web search, and hybrid search conditions.”

May combine two claims.

No severe outliers were identified among the main continuous variables, and their distributional characteristics and homogeneity of variance were generally adequate for the subsequent analyses.

“No severe outliers were identified among the main continuous variables, and their distributional characteristics and homogeneity of variance were generally adequate for the subsequent analyses.”

May combine two claims.

The final analysis included 96 participants, with 32 assigned to each of the GenAI-only, conventional web search, and hybrid search conditions.

“The final analysis included 96 participants, with 32 assigned to each of the GenAI only search, conventional web search, and hybrid search conditions.”

Probably not a substantial claim: may not pass S3 (load-bearing).

The cognitive load measure did not distinguish intrinsic load, extraneous load, or learning-relevant cognitive investment.

“The measure did not distinguish intrinsic load, extraneous load, or learning-relevant cognitive investment.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Distributional characteristics and homogeneity of variance were examined before inferential testing.

“Descriptive statistics were calculated for the principal variables, and distributional characteristics and homogeneity of variance were examined before inferential testing.”

Classified as probably not a substantial claim. May combine two claims.

AOI definitions were established before the primary analyses and applied consistently to all recordings using the same functional coding protocol.

“The AOI definitions were established before the primary analyses and were applied consistently to all recordings using the same functional coding protocol.”

Classified as probably not a substantial claim. May combine two claims.

Fixation duration was accumulated across all page states assigned to the same functional AOI category, and content outside the visible viewport was excluded from dwell time calculations.

“Fixation duration was accumulated across all page states assigned to the same functional AOI category. Content outside the visible viewport was excluded from dwell time calculations.”

Classified as probably not a substantial claim. May combine two claims.

Variables satisfying parametric assumptions were analyzed using one-way ANOVA with post hoc comparisons; when assumptions were not met, the Kruskal-Wallis H test was used; categorical variables used Chi-square or Fisher's exact tests when expected cell frequencies were insufficient.

“Variables satisfying the assumptions of parametric analysis were analyzed using one way analysis of variance (ANOVA), followed by appropriate post hoc comparisons. When these assumptions were not met, the Kruskal-Wallis H test was used. Categorical variables, including the occurrence of cross verification, were analyzed using Chi-square tests or Fisher’s exact tests when expected cell frequencies”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

All eye-tracking sessions were conducted under stable lighting conditions with screen resolution and display scaling held constant.

“All sessions were conducted under stable lighting conditions, with the screen resolution and display scaling held constant (Carter and Luke, 2020; Orquin and Holmqvist, 2018).”

• cites Carter and Luke, 2020 sourceNot checked — no full text of the source
• cites Orquin and Holmqvist, 2018 sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S3 (load-bearing).

Because access to external information differed by condition, source clicks and external-webpage visits were reported as condition-specific descriptive measures.

“Because access to external information differed by condition, source clicks and external-webpage visits were reported as condition-specific descriptive measures.”

May combine two claims.

Cronbach's α for the principal questionnaire scales ranged from 0.81 to 0.88, and Cohen's κ for behavioral coding ranged from 0.82 to 0.91.

“Cronbach’s α coefficients for the principal questionnaire scales ranged from 0.81 to 0.88, while Cohen’s κ values for behavioral coding ranged from 0.82 to 0.91.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

Randomization was performed by a research assistant who was not involved in task scoring, and outcome assessors remained blinded to group allocation during scoring.

“Randomization was performed by a research assistant who was not involved in task scoring, and outcome assessors remained blinded to group allocation during scoring.”

May combine two claims.

Opening a webpage does not establish that its evidence was understood or used; opening a source is not equivalent to verifying it.

“Opening a webpage does not establish that its evidence was understood or used.”

• Also stated in: § Conclusion · plain

  “These results sharpen the distinction between source access and source evaluation: opening a source is not equivalent to verifying it.”

May combine two claims.

The study measured overall perceived cognitive load and did not distinguish intrinsic load, extraneous load, or learning-relevant cognitive investment.

“The study measured overall perceived cognitive load and did not distinguish intrinsic load, extraneous load, or learning-relevant cognitive investment.”

Classified as probably not a substantial claim.

Regression coefficients were interpreted as statistical associations rather than evidence of causal effects.

“Regression coefficients were interpreted as statistical associations rather than evidence of causal effects.”

Classified as probably not a substantial claim. May combine two claims.

All sessions were conducted in the same laboratory using identical monitors, screen resolutions, browser settings, and network configurations.

“To promote comparability, all sessions were conducted in the same laboratory using identical monitors, screen resolutions, browser settings, and network configurations.”

Classified as probably not a substantial claim.

Because the task was rule oriented and carried plausible real-world consequences, the value of verification observed here should not be generalized directly to low-risk, open-ended, or creative tasks.

“Because the task was rule oriented and carried plausible real-world consequences, the value of verification observed here should not be generalized directly to low-risk, open-ended, or creative tasks.”

May combine two claims.

Inter-rater reliability for the information integration and immediate near transfer task scores was ICC = 0.89 and ICC = 0.87, respectively.

“Inter rater reliability for the information integration and immediate near transfer task scores was ICC = 0.89 and ICC = 0.87, respectively.”

Classified as probably not a substantial claim. May combine two claims.

Two coders independently coded behavioral recordings using a standardized scheme; disagreements were resolved through discussion, and intercoder reliability was reported.

“Two coders independently coded the recordings using a standardized scheme; disagreements were resolved through discussion, and intercoder reliability was reported.”

May combine two claims.

When verification requirements are activated and tool affordances permit, AI can support problem framing and information organization while learners retain responsibility for source checking, evidential judgment, and final decisions.

“When verification requirements are activated and tool affordances permit, AI can support problem framing and information organization while learners retain responsibility for source checking, evidential judgment, and final decisions.”

May combine two claims.

Behavioral intention, effort expectancy, facilitating conditions, and other adoption factors were not measured.

“Finally, behavioral intention, effort expectancy, facilitating conditions, and other adoption factors were not measured.”

Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

The study did not measure self-efficacy, need for cognition, epistemic vigilance, or habitual dependence on a particular information-seeking mode; these unmeasured individual differences may influence both tool choice and verification behavior.

“The study also did not measure self-efficacy, need for cognition, epistemic vigilance, or habitual dependence on a particular information-seeking mode; these unmeasured individual differences may influence both tool choice and verification behavior.”

Classified as probably not a substantial claim.

Interpretation therefore relied on converging evidence from gaze, cross-verification, answer revision, and task scores.

“Interpretation therefore relied on converging evidence from gaze, cross-verification, answer revision, and task scores.”

Classified as probably not a substantial claim. May combine two claims.

Eye tracking primarily measures visual attention, while screen recordings and logs capture observable behavior; neither can independently establish comprehension, deep processing, or critical evaluation.

“Eye tracking primarily measures visual attention, while screen recordings and logs capture observable behavior; neither can independently establish comprehension, deep processing, or critical evaluation.”