1. Introduction and problem

In the 3-to-6-year span—kindergarten, preschool, CENDI, early childhood school—a package of three artifacts has been installed as if it counted as early scientific inquiry: a “STEM questions” or 5E-plan chatbot; a sensor/IoT or computer-vision system that classifies “scientific curiosity” or “engagement in the experiment”; and an experiment-script generator from a prompt. All three are visible in coordination and allow a center to exhibit that it “already does science with AI.” The leap—from a generated 5E plan, curiosity labels or a prompt-based script to asserting early scientific inquiry—is authorized neither by evidence nor by science pedagogy in early childhood, where situated practice counts: observation, the child’s genuine question, experimentation with real materials, recording and adult mediation at the science table or corner.

The thesis is restrictive: those artifacts do not constitute early scientific inquiry. At this stage inquiry is situated practice—child hypotheses cultivated at the science table with adults who model, listen, extend and document—; not a curiosity score, engagement dashboard or generated plan that replaces table and adult. Uğraş, Çakır, and Zacharis (2025) survey N = 33 teachers on ChatGPT in science: partial support for resources and plans, with limits of overreliance, misinformation and ethics; not substitution of hands-on inquiry. Liang and Yang (2026) study GenAI in STEM in a Chinese kindergarten (3 classrooms, 6 teachers, 88 children): material facilitation, challenges of child prompts and distracting rhetorical questions. That authorizes asking what was measured: plans or labels, not the practice when a child watches ice melt, asks “why does it disappear?,” tries salt and talks with a mediating adult.

The problem is aggravated by five category confusions that product sheets do not mention and that this article separates rigorously. First: early scientific inquiry is not educational STEM/robotics—although they share “STEM” vocabulary, situated inquiry is not exhausted by kits, programmable sequences or robots as product—; that axis was already treated in the series and here appears only as a limit (Kewalramani, Kidman, and Palaiologou, 2021; Lee, 2026; Qian, 2026). Second: it is not numeracy. Third: it is not emergent literacy. Fourth: it is not orality or narration. Fifth: it is not generic academic formative assessment—although there is teacher observation, the construct is not an automated curiosity score. This work does not recycle articles on STEM/robotics, numeracy, literacy, orality, executive functions, formative assessment, motor development, artistic creativity, interactions, documentation, SEL, free play, UDL, participation, planning or teacher training. The question is early scientific inquiry: what counts as such when an early childhood center “does AI and science.”

Ramanathan, Carter, and Wenner (2022) anchor a preschool inquiry framework—observe–ask–explore–record–communicate with mediation—; NAEYC (2022) and OECD (2021, 2023) require meaningful interactions and subordinated digitalization. Pedagogical inference: inquiry is not fulfilled by generating plans or algorithmically scoring curiosity. The contributions are three: to reconstruct the state of the art separating early scientific inquiry (observation; question; hypothesis; experimentation; recording; mediation at the science table) from the STEM/5E chatbot, curiosity/engagement scoring and prompt experiment script; to examine three families of empirical cases; and to offer four tests for deciding when a kindergarten may assert that early scientific inquiry exists, and not merely a chatbot, classifier or generator.

2. State of the art: from situated inquiry practice to the exhibited artifact

It is useful to separate four strata that the market for “AI for early scientific inquiry” usually mixes. The first is the construct of early scientific inquiry for ages 3 to 6 as practice of observation, question, exploration with real materials, recording and communication in everyday classroom contexts (Ramanathan, Carter, and Wenner, 2022; Yildiz and Guler Yildiz, 2021; NAEYC, 2022; OECD, 2021). The second is the pedagogical craft that cultivates it—science table or corner, hands-on experimentation, conversation about findings, teacher mediation that listens to child questions and extends hypotheses— (Ramanathan, Carter, and Wenner, 2022; Moffit, 2025, with transfer marked; NAEYC, 2022; OECD, 2021, 2023). The third is evidence on teacher chatbots, GenAI in STEM, AI affordances in ECE and AI literacy—without equating it to situated inquiry— (Uğraş, Çakır, and Zacharis, 2025; Liang and Yang, 2026; Chen, 2024; Su and Yang, 2022; Su and Zhong, 2022; Ljungcrantz, 2026; Nikolopoulou, 2025; Qian, 2026; Lee, 2026; Kewalramani, Kidman, and Palaiologou, 2021). The fourth is the framework of rights, systems and developmentally appropriate practice, which treats the 3–6-year-old as a subject of situated inquiry, not as a signal vector for a scoring pipeline nor as a passive receiver of generated plans (UNESCO, 2021; Miao and Holmes, 2023; European Commission, 2022; U.S. Department of Education, 2023).

In construct and craft, Ramanathan, Carter, and Wenner (2022) propose a preschool inquiry framework—observation, open questions, guided exploration, recording and communication—: framework, not 5E plan by chatbot. Yildiz and Guler Yildiz (2021) report an association between creativity and scientific process skills in preschool: empirical finding of process craft, not app score. Moffit (2025) documents inquiry with animals in kindergarten—transfer marked by integrated literacy. NAEYC (2022) and OECD (2021, 2023) anchor active exploration and meaningful interactions. Inference: a chatbot does not observe with the child; an adult at the science table does.

In artifacts, Uğraş, Çakır, and Zacharis (2025) report chatbot as partial support (N = 33 teachers), not child inquiry. Liang and Yang (2026) report GenAI in STEM (3 classrooms, 88 children) with inappropriate outputs and rhetorical distraction. Chen (2024), Su and Yang (2022), Su and Zhong (2022) and Ljungcrantz (2026) map AI in ECE without substituting craft. Kewalramani, Kidman, and Palaiologou (2021) study inquiry literacy with AI toys—a limit against STEM/robotics. Qian (2026) and Lee (2026) develop AI literacy for ages 4–6 and pre-K: construct distinct from inquiry. In systems, UNESCO (2021), Miao and Holmes (2023), European Commission (2022) and U.S. Department of Education (2023) require human oversight and pedagogical validation. Kharbanda and Khunyakari (2025) analyze design in narrative contexts—transfer marked—as a limit against confusing narration with inquiry.

3. Review method

A critical narrative review was conducted, not a primary meta-analysis. The purpose was not to estimate a homogeneous effect size of STEM chatbots, curiosity classifiers or script generators, but to articulate a pedagogical-category argument with verified sources. Inclusion criteria: (a) 2021–2026, with transfer explicitly marked when the sample or level does not equate to kindergarten ages 3–6; (b) early scientific inquiry, scientific thinking, scientific process skills, science education in early childhood, generative AI/chatbots in early science or mappings of AI in ECE; (c) relevance for kindergarten, preschool, CENDI or ages 3–6; (d) peer-reviewed journal, DOI or report from NAEYC, UNESCO, OECD, European Commission or education department; (e) verifiable DOI or editorial page. Excluded as central object were axes already used in this series—educational STEM/robotics as its own article, numeracy, emergent literacy, orality/narration, academic formative assessment, executive functions, motor/FMS, visual artistic creativity, CLASS/serve-and-return interactions, pedagogical documentation, SEL as substitute programme, generic free play, generative tutoring, gaps, privacy, UDL, family–school, continuous training, participation and planning—although some appear as category limits.

The search was executed on 1 September 2026 (slot 21:02 America/Mexico_City) on DOI pages, Crossref, Springer, Taylor & Francis, MDPI, Elsevier, JAIR, OECD iLibrary, UNESDOC, NAEYC and editorial sites. Each source was verified against at least one of those pages. Empirical finding, conceptual or normative framework, and pedagogical inference marked as such were distinguished. Priority was given to the distinction early scientific inquiry / STEM-robotics / numeracy / literacy / orality / formative assessment, and to the caution of not translating generated 5E plans, curiosity scores or dashboard-reported engagement into pedagogy of the inquiry table and the child’s genuine question. N, d, r, AUC and DOI were not invented: when an artifact (sensor/IoT, computer vision of curiosity) lacks a verified study in the corpus, it is discussed as a category ceiling supported by general AI-in-ECE mappings (Chen, 2024; Su and Yang, 2022; Ljungcrantz, 2026), not as an invented empirical finding.

4. Case 1. STEM/5E chatbot, curiosity/engagement scoring or prompt experiment scripts do not constitute early scientific inquiry

Uğraş, Çakır, and Zacharis (2025), in Computers, survey N = 33 early childhood teachers in Türkiye on ChatGPT in science: partial utility for planning and resources; limits of overreliance, misinformation and ethics. Empirical finding of teacher support, not child inquiry. Inference: the gesture “5E plan from chatbot = inquiry session” is a ceiling of generative planning. Ramanathan, Carter, and Wenner (2022) and NAEYC (2022) ask for observation, genuine question and real materials at the science table. A useful chatbot plan can coexist with absence of the child’s hands-on inquiry.

Liang and Yang (2026), in Early Education and Development, study GenAI in STEM in a Chinese kindergarten (3 classrooms, 6 teachers, 88 children): partial material facilitation; challenges of child prompts, inappropriate outputs and distracting rhetorical questions. Empirical finding of digital teacher support, not substitution of table or mediation. Miao and Holmes (2023) require pedagogical validation of generative AI. Inference: a prompt script may be teacher preparation; inquiry begins with hands in water, ice or magnets and a mediating adult.

The sensor/IoT or computer vision of “curiosity”/“engagement” lacks in the corpus a verified study reporting N, d, r or AUC in ages 3–6; figures are not invented. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) map analytics and sensors as a trend, without equivalence to inquiry. Inference: classifying signals ≠ declaring inquiry; the dashboard may rise without documentation of question or evidence. Nikolopoulou (2025), UNESCO (2021) and U.S. Department of Education (2023) require a child-centered approach and human oversight. The three artifacts substitute listening, real materials or pedagogical observation with plan, script or label.

5. Case 2. What the kindergarten does when early scientific inquiry exists: observation, question, experimentation and mediation at the science table

Ramanathan, Carter, and Wenner (2022) define preschool inquiry: sustained observation, open questions, guided exploration, multimodal recording and communication of findings. Usable framework: observe, ask “what will happen if…?,” manipulate real materials and converse at table, yard or water tray—not a printed 5E plan. Yildiz and Guler Yildiz (2021) report creativity–scientific process skills association: craft of observing, comparing, predicting and communicating, not curiosity feed. Moffit (2025) documents “Head Zoologist” in kindergarten—transfer marked by integrated literacy—: mediated observation, recording and communication.

NAEYC (2022) and OECD (2021, 2023) anchor active exploration, meaningful interactions and subordinated digitalization. Teacher mediation—NAEYC, 2022; OECD, 2021— is organizing materials, listening to questions, modeling vocabulary (“float,” “melt”), documenting and extending without imposing answers from a generated script. Kharbanda and Khunyakari (2025) show inquiry in narrative contexts—transfer marked—: conversation about findings is craft, not pure orality.

6. Case 3. Early scientific inquiry is not STEM/robotics, numeracy, literacy, orality or formative assessment

The first category boundary is educational STEM/robotics. A kit, an AI-interfaced robot or an AI literacy curriculum may coexist in a center; they do not alone constitute early scientific inquiry of observation, question and experimentation with real materials at ages 3–6. Kewalramani, Kidman, and Palaiologou (2021) study AI robotic toys and inquiry literacy: finding of mediated inquiry literacy, not equivalence “install robot = scientific inquiry.” Lee (2026) proposes PL-AI, play-centered AI literacy for pre-K and kindergarten: finding of play-based design for AI literacy; inference: it does not replace the science table. Qian (2026) develops an AI literacy pedagogical framework for ages 4–6 with dialogic scaffolding: AI literacy framework, not scientific inquiry. Liang and Yang (2026) report GenAI in STEM projects: limit inference—STEM as school project ≠ situated inquiry at the science table when hands-on disappears. This article does not recycle the STEM/robotics piece: it uses it only as a boundary.

The second boundary is numeracy. Counting objects in an experiment may be part of inquiry; it does not convert early scientific inquiry into numeracy teaching as the central construct (Ramanathan, Carter, and Wenner, 2022; Yildiz and Guler Yildiz, 2021). The third is emergent literacy: Moffit (2025) integrates inquiry with literacy—transfer marked—; recording by drawing is evidence, not literacy as central object. The fourth is orality/narration: Kharbanda and Khunyakari (2025) show inquiry in stories—transfer marked—; telling without manipulating is not equivalent to experimenting. The fifth is automated formative assessment: observing inquiry is professional judgment; not an algorithmic curiosity score. Chen (2024), Su and Yang (2022), Ljungcrantz (2026), Nikolopoulou (2025) and Miao and Holmes (2023) confirm that AI in ECE must be subordinated to the adult at the science table—material preparation or professional reflection—, not as an autonomous 5E chatbot, generator that replaces real materials or classifier declaring curiosity fulfilled.

7. Inferential framework: four tests for asserting early scientific inquiry, not an artifact

The following framework is pedagogical inference of this article, anchored in the cases and verified instruments. It is not a new international standard. It distinguishes four tests. If a kindergarten, preschool, CENDI or early childhood school does not pass them, it cannot declare that a STEM/5E chatbot, curiosity or engagement sensor/CV, or experiment-script generator constitutes early scientific inquiry.

7.1. Test of situated practice of observation, question and experimentation with real materials, not the generated 5E plan. Ramanathan, Carter, and Wenner (2022) and Yildiz and Guler Yildiz (2021) define craft as mediated observation, question, exploration and recording. Uğraş, Çakır, and Zacharis (2025) report chatbot as partial plan support (N = 33 teachers), not child inquiry. Inference: inquiry evidence is verified in whether the child observed, asked and manipulated real materials. If the center’s “evidence” is a printed 5E plan or on-screen script, the center has done generative planning, not early scientific inquiry.

7.2. Test of the science table, evidence recording and conversation about findings, not the prompt experiment script. NAEYC (2022) places active environmental exploration in developmentally appropriate practice. Miao and Holmes (2023) require pedagogical validation of generative AI. Liang and Yang (2026) report inappropriate outputs and rhetorical questions distracting from hands-on (3 classrooms; 6 teachers; 88 children). Inference: producing a prompt script does not demonstrate inquiry table, recording or mediated conversation. A script may exist; it does not sign the practice.

7.3. Test of teacher mediation and pedagogical observation of inquiry, not the curiosity or engagement dashboard. OECD (2021, 2023) require meaningful interactions and subordinated digitalization. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) map sensors and analytics as emerging affordances, without equivalence to inquiry. Inference: “high curiosity” or “high engagement” on a dashboard may raise a threshold without raising mediation quality. Pedagogical observation reads the child’s question and experiment evidence; the dashboard counts a label.

7.4. Test of category distinction and professional judgment, not the product catalog. Early scientific inquiry ≠ STEM/robotics, numeracy, literacy, orality or automated formative assessment. Kewalramani, Kidman, and Palaiologou (2021), Qian (2026) and Lee (2026) prevent confusing inquiry literacy or AI literacy with situated inquiry. Moffit (2025) shows overlap with literacy without identity of constructs—transfer marked. UNESCO (2021), Miao and Holmes (2023), European Commission (2022), U.S. Department of Education (2023), NAEYC (2022) and OECD (2021, 2023) require human oversight, pedagogical validation and not substituting professional judgment. Inference: a center cannot treat the infant as a signal emitter for scoring nor as a passive receiver of generated plans. Early scientific inquiry is not fulfilled by algorithmically scoring curiosity. It is fulfilled by practicing observation, genuine question, experimentation with real materials, recording and conversation about findings with adults who mediate at the science table.

The framework admits digital tools when subordinated to teacher preparation and validated professional observation (Uğraş, Çakır, and Zacharis, 2025, as partial support with limits; Liang and Yang, 2026, as cautious material facilitation). It rejects declaring early scientific inquiry through an autonomous 5E-plan chatbot, script generator that replaces real materials or curiosity/engagement classifier (Miao and Holmes, 2023; Nikolopoulou, 2025).

The four tests are read together. Passing only the first—there were materials—without mediation or category distinction is not enough. Passing only the fourth—the center distinguishes constructs—without situated practice neither.

8. Discussion

Three tensions organize the discussion. The first is between exhibiting a planning, generation or scoring artifact and exercising early scientific inquiry. It is a finding that N = 33 teachers perceive partial ChatGPT utility for resources and plans with overreliance and ethics limits (Uğraş, Çakır, and Zacharis, 2025); that GenAI facilitates STEM materials in 3 classrooms with 88 children but introduces distracting rhetorical questions (Liang and Yang, 2026); and that AI-in-ECE mappings grow without equivalence to situated inquiry (Chen, 2024; Su and Yang, 2022; Ljungcrantz, 2026). It is framework that early scientific inquiry for ages 3–6 is played out in observation, question, experimentation with real materials and mediation at the science table (Ramanathan, Carter, and Wenner, 2022; Yildiz and Guler Yildiz, 2021; NAEYC, 2022). It is not a finding that chatbot, generator or dashboard produce the craft the inquiry table requires. The three artifacts plan, generate or label what engineering can deliver and declare what only mediated practice would authorize.

The second is between automated curiosity/engagement assessment and everyday science pedagogy. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) objectify analytics and sensor affordances as a trend; they do not report in the corpus verified studies validating curiosity scoring in preschool ages 3–6—therefore metrics are not invented. Inference: insisting that the kindergarten “already has scientific inquiry” because a model classifies sufficient engagement or curiosity is inverted pedagogy. An algorithmic proxy is made to stand for observation, genuine question and hands-on experimentation. The infant becomes a signal emitter for the classifier; the adult, dashboard supervisor.

The third is between generated script and integral inquiry craft. Miao and Holmes (2023) and Nikolopoulou (2025) set limits for child-centered GenAI; Liang and Yang (2026) document inappropriate outputs and rhetorical distraction. Inference: selling a prompt script as “inquiry with AI” confuses textual product with situated practice. An autonomous chatbot because the system needs output does not replace the adult who pours water, listens to “does it float or sink?” and records the child’s hypothesis with pencil and paper.

Additional inference: the artifacts share a visibility economy—plan, label, script—versus inquiry verified in everyday classroom life (OECD, 2021; NAEYC, 2022). Qian (2026), Lee (2026) and Kewalramani, Kidman, and Palaiologou (2021) confirm distinct constructs. Miao and Holmes (2023), UNESCO (2021) and European Commission (2022) subordinate AI to professional judgment.

The four tests in section 7 allow these tensions to be read with operational criteria. When a kindergarten, preschool or CENDI coordinator evaluates whether “early scientific inquiry with AI” exists, it is useful to apply them in sequence, not in isolation. Test 7.1 requires evidence that the child observed, asked and manipulated real materials—not that a printed 5E plan or on-screen script exists (Ramanathan, Carter, and Wenner, 2022; Uğraş, Çakır, and Zacharis, 2025). Test 7.2 asks whether there was a science table, evidence recording and conversation about findings—not only a prompt-generated script (NAEYC, 2022; Liang and Yang, 2026). Test 7.3 distinguishes teacher mediation and pedagogical observation from a curiosity or engagement dashboard (OECD, 2021, 2023; Chen, 2024). Test 7.4 prevents confusing inquiry with STEM/robotics, numeracy, literacy, orality or automated formative assessment (Kewalramani, Kidman, and Palaiologou, 2021; Qian, 2026; Lee, 2026). Pedagogical inference: a center that passes only the fourth test—knows how to name constructs—but fails the first—no situated practice—exhibits AI literacy or inquiry literacy, not declared early scientific inquiry.

The most illustrative empirical contrast in the corpus does not come from the artifacts but from documented craft when real inquiry exists. Ramanathan, Carter, and Wenner (2022) describe a framework of observation, open questions, guided exploration, recording and communication in preschool: practice a chatbot cannot perform because it does not share the table with the child. Yildiz and Guler Yildiz (2021) link creativity and scientific process skills in preschool—observe, compare, predict, communicate—without algorithmic scores. Moffit (2025), with transfer marked by integrated literacy, documents the “Head Zoologist” kindergarten case: animal observation, recording and adult-mediated communication—not a script generator or curiosity classifier. Kharbanda and Khunyakari (2025), also with transfer marked, show inquiry in narrative contexts: conversation about findings is science craft, not pure orality or generated script. These cases define the minimum floor the three artifacts do not reach on their own.

The partial utility Uğraş, Çakır, and Zacharis (2025) report—N = 33 early childhood teachers in Türkiye—deserves careful reading. The finding is not that ChatGPT produces child inquiry, but that some teachers perceive it useful for planning and resources, with explicit limits of overreliance, misinformation and ethics. That places the STEM/5E chatbot in the stratum of subordinated teacher support that NAEYC (2022), OECD (2021, 2023) and Miao and Holmes (2023) require: a preparation tool, not a substitute for table or mediation. Liang and Yang (2026) confirm the ambivalence: GenAI facilitates materials in 3 classrooms with 6 teachers and 88 children, but introduces inappropriate outputs and rhetorical questions distracting from hands-on work. Inference: partial utility does not authorize the declaration “we already guarantee early scientific inquiry”; it authorizes, at most, “we use AI to prepare resources with caution and human oversight.”

On the systems and rights axis, UNESCO (2021), European Commission (2022) and U.S. Department of Education (2023) converge on requiring human oversight, pedagogical validation and not substituting professional judgment. Nikolopoulou (2025) warns about child-centered GenAI integration: promises and challenges not resolved by installing an engagement classifier. Su and Zhong (2022) propose AI curriculum design in early childhood as a future direction—AI literacy curriculum, not early scientific inquiry. Su and Yang (2022) and Chen (2024) map emerging affordances without equivalence to situated inquiry. Ljungcrantz (2026) confirms growth of AI–ECE interaction as a field trend, not as evidence that a sensor/IoT or computer-vision model validates scientific curiosity in CENDI ages 3–6. Inference: institutional policies demanding “inquiry with AI” without the four tests invert the normative hierarchy: the exhibited product commands over the craft NAEYC (2022) and OECD (2021) define.

Category confusion has concrete consequences in classroom life. When a kindergarten declares inquiry because it deployed a “STEM questions” chatbot, the adult may stop listening to the child’s genuine question—“why does the ice disappear?”—and deliver instead a rhetorical question from the generated script, as Liang and Yang (2026) warn. When a preschool exhibits a curiosity dashboard, coordination may reward algorithmic thresholds instead of documenting evidence in drawing, photo or conversation, as Ramanathan, Carter, and Wenner (2022) ask. When a CENDI replaces real materials with an experiment script, the water tray, magnet, salt and child hypothesis that Yildiz and Guler Yildiz (2021) associate with scientific process skills disappear. Inference: early scientific inquiry erodes not by using AI at all, but by treating planning, generation or scoring artifacts as if they were the practice itself.

Qian (2026) and Lee (2026) offer an additional pedagogical limit. Qian (2026) develops an AI literacy framework for ages 4–6 with dialogic scaffolding; Lee (2026) proposes PL-AI, a play-centered AI literacy curriculum for pre-K and kindergarten. Both constructs—AI literacy—are legitimate and distinct from early scientific inquiry. Kewalramani, Kidman, and Palaiologou (2021) study inquiry literacy with AI robotic toys: mediated inquiry literacy, not equivalence “install robot = scientific inquiry.” Inference: a center may advance in AI literacy or inquiry literacy and still not fulfill early scientific inquiry if there is no sustained observation, genuine question, experimentation with real materials, recording and adult mediation at the science table. Confusing digital literacies with situated inquiry reproduces the same invalid leap as confusing a generated 5E plan with a hands-on science session.

Finally, the discussion recognizes a legitimate space for subordinated digital tools. Uğraş, Çakır, and Zacharis (2025) and Liang and Yang (2026) do not proscribe GenAI; they document partial support with limits. Miao and Holmes (2023) and Nikolopoulou (2025) ask for pedagogical validation and a child-centered approach. OECD (2023) situates infants in the digital age without turning them into signal vectors for scoring pipelines. Inference: the chatbot may help prepare materials for the science table; the generator may propose variations the adult will adapt; neither replaces shared observation, the child’s question, manipulation of ice or magnets, pencil recording and mediated conversation that Moffit (2025), Ramanathan, Carter, and Wenner (2022) and NAEYC (2022) describe as early scientific inquiry craft. What cannot be done is to invert the sequence: install artifact, exhibit dashboard, print 5E plan or deploy script, and declare inquiry fulfilled. Inquiry is verified at the table, with real materials and a present adult—not on the classifier screen nor in the chatbot output.

9. Limits

This review is narrative. It does not apply PRISMA nor estimate combined primary effects. Uğraş, Çakır, and Zacharis (2025) survey N = 33 teachers, not direct child inquiry nor chatbot versus mediated science-table comparison. Liang and Yang (2026) study 3 classrooms, 6 teachers and 88 children in a Chinese kindergarten: case study, not randomized trial of GenAI versus inquiry without AI. Ramanathan, Carter, and Wenner (2022) propose a conceptual framework, not an AI intervention. Yildiz and Guler Yildiz (2021) correlate creativity and process skills in preschool, not chatbots or sensors. Moffit (2025) is a kindergarten case with integrated literacy: transfer marked. Kharbanda and Khunyakari (2025) analyze design in narrative contexts: transfer marked. Kewalramani, Kidman, and Palaiologou (2021) study inquiry literacy with AI toys, not a classic science table. Qian (2026) and Lee (2026) address AI literacy, not scientific inquiry. Chen (2024), Su and Yang (2022), Su and Zhong (2022), Ljungcrantz (2026) and Nikolopoulou (2025) map AI in ECE, not early scientific inquiry as primary variable with curiosity scoring. No verified corpus trials of computer vision or IoT scoring scientific curiosity in CENDI ages 3–6 were located; the second artifact is discussed as a category ceiling, not an invented finding. NAEYC, UNESCO and OECD are framework sources. No Latin American trials comparing adult-mediated inquiry versus autonomous STEM chatbot in kindergarten were located. Section 7 inferences are pedagogical-category hypotheses, not implementation evidence.

10. Conclusions

A chatbot that proposes “STEM questions” or 5E plans, a sensor/IoT or computer-vision model that classifies “scientific curiosity” or “engagement in the experiment,” or a generator that produces experiment scripts from a prompt do not constitute early scientific inquiry in an early childhood center. Verified evidence does not authorize that declaration. Uğraş, Çakır, and Zacharis (2025) report partial perceived ChatGPT support among N = 33 early childhood teachers, with overreliance, misinformation and ethics limits—generative planning, not child inquiry. Liang and Yang (2026) document GenAI in 3 classrooms, 6 teachers and 88 children with child-prompt, inappropriate-output and distracting rhetorical-question challenges—material facilitation, not hands-on substitution. Chen (2024), Su and Yang (2022) and Ljungcrantz (2026) confirm AI affordances in ECE without equivalence to situated inquiry. When early scientific inquiry exists in early childhood, there is situated practice: preschool inquiry framework (Ramanathan, Carter, and Wenner, 2022); scientific process skills associated with creativity (Yildiz and Guler Yildiz, 2021); inquiry-with-animals kindergarten case—transfer marked— (Moffit, 2025); active environmental exploration and developmentally appropriate practice (NAEYC, 2022; OECD, 2021, 2023). Early scientific inquiry is distinguished from STEM/robotics, numeracy, emergent literacy, orality/narration and automated formative assessment. Kewalramani, Kidman, and Palaiologou (2021), Qian (2026) and Lee (2026) confirm distinct constructs—inquiry literacy and AI literacy are not declared inquiry. Current guidelines require human oversight, pedagogical validation and not substituting professional judgment (UNESCO, 2021; Miao and Holmes, 2023; European Commission, 2022; U.S. Department of Education, 2023).

Where sources do not measure a kindergarten, this article does not assert it. Where they measure chatbots, GenAI or AI mappings, it does not translate them into pedagogical early scientific inquiry. Accompanying three- to six-year-olds in science is exercising observation, listening to genuine questions, experimenting with real materials, recording evidence and conversing about findings with adult mediation at the science table or corner. The rest is a generated 5E plan, prompt script and algorithmic classification of curiosity or engagement. It is not early scientific inquiry in early childhood education, and it must not be presented as what it is not.

Laboratorio Editorial de NEXTECH.IA / Ingeniero Mitre.

References

  1. Chen, J. J. (2024). A scoping study on AI affordances in early childhood education: Mapping the global landscape, identifying research gaps, and charting future research directions. Journal of Artificial Intelligence Research, 81, 701–740. https://doi.org/10.1613/jair.1.16882
  2. European Commission. (2022). Ethical guidelines on the use of artificial intelligence (AI) and data in teaching and learning for educators. Publications Office of the European Union. https://education.ec.europa.eu/focus-topics/digital-education/action-plan
  3. Kewalramani, S., Kidman, G., y Palaiologou, I. (2021). Using Artificial Intelligence (AI)-interfaced robotic toys in early childhood settings: a case for children’s inquiry literacy. European Early Childhood Education Research Journal, 29(5), 652–668. https://doi.org/10.1080/1350293x.2021.1968458
  4. Kharbanda, D., y Khunyakari, R. (2025). Analyzing Young Children’s Thinking on Design Problems Embedded in Story Contexts. Early Childhood Education Journal. https://doi.org/10.1007/s10643-023-01551-2
  5. Lee, J. (2026). Play with AI (PL-AI): A play-centered, design-based curriculum for AI literacy in pre-K and kindergarten. Computers and Education: Artificial Intelligence. https://doi.org/10.1016/j.caeai.2026.100569
  6. Liang, J., y Yang, W. (2026). Using Generative Artificial Intelligence to Facilitate Early STEM Learning: A Case Study in a Chinese Kindergarten. Early Education and Development. https://doi.org/10.1080/10409289.2026.2669922
  7. Ljungcrantz, L. (2026). The interaction of AI and early childhood education. A state-of-the-art review 2020–2024. Early Childhood Education Journal, 54, 3565–3581. https://doi.org/10.1007/s10643-025-02079-3
  8. Miao, F., y Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535
  9. Moffit, C. A. (2025). “I’m the Head Zoologist!”: Integrating Science and Literacy Learning in Kindergarten with Inquiry and Animals. Early Childhood Education Journal, 53(2), 529–538. https://doi.org/10.1007/s10643-023-01587-4
  10. NAEYC. (2022). Developmentally appropriate practice in early childhood programs serving children from birth through age 8 (4.ª ed.). NAEYC. https://www.naeyc.org/resources/pubs/books/dap-fourth-edition
  11. Nikolopoulou, K. (2025). Child-centered integration of generative AI in early learning: Balancing promises and challenges. AI, Brain and Child. https://doi.org/10.1007/s44436-025-00023-1
  12. OECD. (2021). Starting Strong VI: Supporting meaningful interactions in early childhood education and care. OECD Publishing. https://doi.org/10.1787/f47a06ae-en
  13. OECD. (2023). Empowering young children in the digital age (Starting Strong). OECD Publishing. https://doi.org/10.1787/50967622-en
  14. Qian, Y. (2026). Building early AI literacy: developing a pedagogical framework for early childhood education. AI, Brain and Child. https://doi.org/10.1007/s44436-026-00030-w
  15. Ramanathan, G., Carter, D., y Wenner, J. (2022). A Framework for Scientific Inquiry in Preschool. Early Childhood Education Journal, 50(7), 1263–1277. https://doi.org/10.1007/s10643-021-01259-1
  16. Su, J., y Yang, W. (2022). Artificial intelligence in early childhood education: A scoping review. Computers and Education: Artificial Intelligence, 3, 100049. https://doi.org/10.1016/j.caeai.2022.100049
  17. Su, J., y Zhong, Y. (2022). Artificial Intelligence (AI) in early childhood education: Curriculum design and future directions. Computers and Education: Artificial Intelligence, 3, 100072. https://doi.org/10.1016/j.caeai.2022.100072
  18. U.S. Department of Education, Office of Educational Technology. (2023). Artificial intelligence and the future of teaching and learning: Insights and recommendations. U.S. Department of Education. https://www.ed.gov/sites/ed/files/documents/ai-report/ai-report.pdf
  19. Uğraş, M., Çakır, Z., y Zacharis, G. (2025). ChatGPT in Early Childhood Science Education: Can It Offer Innovative Effective Solutions to Overcome Challenges? Computers, 14(9), 368. https://doi.org/10.3390/computers14090368
  20. UNESCO. (2021). Recommendation on the ethics of artificial intelligence. UNESCO. https://unesdoc.unesco.org/ark:/48223/pf0000381137
  21. Yildiz, C., y Guler Yildiz, T. (2021). Exploring the relationship between creative thinking and scientific process skills of preschool children. Thinking Skills and Creativity, 39, 100795. https://doi.org/10.1016/j.tsc.2021.100795