1. Introduction and problem

In the 3–6 age range—nursery, preschool, CENDI, infant school—a package of four artifacts has been installed that purport to stand in for environmental education or ECEfS: a chatbot of “eco-tips” or “green classroom” plans; an app that scores “ecological behavior” or “green footprint”; a computer vision model that classifies “sustainable acts” in the yard; and a generator of environmental education units from a prompt. All four are visible in coordination and allow the center to display that it “already does environmental education with AI.” The leap—from tip, score, label, or unit to claiming ECEfS—is not authorized by evidence from AI mappings in ECE nor by environmental pedagogy when there is care of the nearby environment, sensory exploration, everyday habits, and adult mediation.

The thesis of this article is restrictive. A chatbot that gives “eco-tips” or “green classroom” plans, an app that scores a child’s “ecological behavior” or “green footprint,” a computer vision model that classifies “sustainable acts” in the yard, or a generator that produces environmental education units from a prompt do not constitute environmental education / sustainability in early childhood. At this stage, ECEfS is situated practice of care of the nearby environment, sensory exploration of nature, everyday habits—water, waste, yard, garden—, ethical conversation with adult mediation, and place belonging; not a green score nor a generated plan that replaces outdoor experience and the adult. Ardoin and Bowers (2020) anchor in systematic review that ECEE is associated with environmental, cognitive, and socioemotional outcomes under play-based and nature-rich pedagogies: synthesis of craft, not of chatbot. Ernst, McAllister, Siklander and Storli (2021) link nature play with Education for Sustainability across 32 studies. Lepičnik Vodopivec and Šindić (2025) compare eco-programs versus no eco-program (N = 114 teachers). Watt and Frydenberg (2024) document Learning in Nature (Melbourne; 2021/2022) with pro-environmental outcomes and SEL as a lateral outcome—not the axis of this article. That authorizes asking what was measured: tip, point, or label, not the practice when a child waters the garden, sorts waste, explores the yard, and converses about water.

The problem is aggravated by five category confusions that product sheets do not mention and that this article separates with rigor. First: ECEfS / early environmental education is not early scientific inquiry—although observing an insect may touch both constructs, the environmental craft is not exhausted in a science table or experimental hypotheses; that axis was already treated in the series and here appears only as a limit. Second: it is not STEM nor educational robotics. Third: it is not SEL as a generic socioemotional competencies program—Watt and Frydenberg (2024) show SEL as a lateral outcome of a nature program; they do not authorize selling SEL as if it were ECEfS. Fourth: it is not green score nor gamification of “ecological behavior.” Fifth: it is not generative curricular planning—although an adult may use AI to prepare materials, the unit-by-prompt does not replace the yard, the garden, nor mediation. This work does not recycle articles on early scientific inquiry, STEM/robotics, numeracy, literacy, orality, free play, UDL, inclusion, SEL as main theme, documentation, participation, planning, executive functions, creativity, music, motor skills, joint attention, interactions, or teacher training. The question is of environmental education / sustainability in early childhood: what counts as such when a center “does AI and green classroom.”

There is, moreover, a coordination economy: an eco-tip, footprint score, sustainable-act label, and printed unit fit on one slide; yard care and ethical conversation about water do not. Ardoin and Bowers (2020), Ernst et al. (2021), Borg and Pramling Samuelsson (2022), and NAEYC (2022) situate the craft in the nearby environment; OECD (2021, 2023) demand meaningful interactions and subordinate digitalization. Pedagogical inference: ECEfS is not fulfilled by generating tips or scoring footprint. The contributions are three: separate ECEfS from the artifact package; examine three families of cases; and offer four tests to decide when a nursery can claim environmental education / sustainability, and not only tip, score, vision, or prompt.

2. State of the art: from situated care practice to the artifact on display

It is useful to separate four strata that the market of “AI for environmental education in early childhood” usually mixes. The first is the construct of ECEE / ECEfS for ages 3–6 as practice of care of the nearby environment, nature-rich exploration, everyday habits, and place belonging (Ardoin and Bowers, 2020; Ernst et al., 2021; Borg and Pramling Samuelsson, 2022; Eilam, 2023, with TRANSFER marked when age is mixed). The second is the pedagogical craft that cultivates it—nature play, eco-programs, child agency, teacher implementation, water/waste/yard/garden habits, ethically mediated conversation—(Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér, 2021; Lepičnik Vodopivec and Šindić, 2025; Watt and Frydenberg, 2024; Šindić, Višnjić-Jevtić and Lepičnik Vodopivec, 2025; Borg and Gericke, 2021; NAEYC, 2022; OECD, 2021, 2023). The third is evidence of AI affordances in ECE, GenAI reviews, and teacher co-pilot frameworks—without equating them to situated ECEfS—(Chen, 2024; Su and Yang, 2022; Su and Zhong, 2022; Ljungcrantz, 2026; Nikolopoulou, 2025; Zhu, 2025). The fourth is the framework of rights, systems, and developmentally appropriate practice, which treats the 3–6 child as a subject of situated care of place, not as a vector of signals for a scoring pipeline nor as a passive receiver of generated eco-tips (UNESCO, 2021; Miao and Holmes, 2023; U.S. Department of Education, 2023).

In construct and craft, Ardoin and Bowers (2020) synthesize ECEE with environmental, cognitive, and socioemotional outcomes under play-based and nature-rich pedagogies (2020 anchor). Pedagogical inference: the craft is nature-rich and mediated, not footprint score. Ernst et al. (2021) link nature play and sustainability across 32 studies. Borg and Pramling Samuelsson (2022) analyze child agency and holistic EfS under Lpfö2018. Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021) document teacher implementation of EfS. Lepičnik Vodopivec and Šindić (2025) compare eco-programs versus no eco-program (N = 114 teachers). Watt and Frydenberg (2024) report pro-environmental outcomes and lateral SEL in Learning in Nature (Melbourne; 2021/2022). Šindić, Višnjić-Jevtić and Lepičnik Vodopivec (2025) situate children as agents of change. Borg and Gericke (2021) teach local/global social sustainability. Eilam (2023) articulates place-based education—TRANSFER marked. NAEYC (2022) and OECD (2021, 2023) anchor exploration and meaningful interactions. Pedagogical inference: a chatbot does not water the garden; an adult in the yard does.

In the artifact stratum, Chen (2024), Su and Yang (2022), Su and Zhong (2022), and Ljungcrantz (2026) map affordances, curricula, and state of the art of AI in ECE without equating them to care of the environment nor to nature play. Nikolopoulou (2025) balances promises and challenges of child-centered GenAI. Zhu (2025) proposes a conceptual framework of AI as Co-Pilot for sustainability education in ECE with emphasis on empowerment without replacement: contrast framework—teacher co-pilot ≠ eco-tips chatbot for the child nor green score. In the systems stratum, UNESCO (2021) demands human supervision and AI ethics. Miao and Holmes (2023) set pedagogical validation and age thresholds for generative AI—direct framework for the environmental units-by-prompt generator. U.S. Department of Education (2023) demands that AI support, not replace, professional judgment.

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 eco-tips chatbots, green-footprint apps, sustainable-act classifiers, or unit generators, but to articulate a pedagogical category argument with verified sources. Inclusion criteria: (a) 2021–2026, with Ardoin and Bowers (2020) accepted as foundational systematic review anchor, and with TRANSFER explicitly marked when the sample or level does not equate to nursery 3–6; (b) ECEE, ECEfS, nature play, eco-programs, place-based education, sustainability in preschool, or AI in early childhood education with relevance for the artifact/craft contrast; (c) relevance for nursery, preschool, CENDI, or ages 3–6; (d) peer-reviewed journal, DOI, or NAEYC, UNESCO, OECD, or department of education report; (e) verifiable DOI or publisher page. Excluded as central object were axes already used in this series—early scientific inquiry, STEM/robotics, numeracy, literacy, orality, generic free play, UDL, inclusion, SEL as main program, pedagogical documentation, participation, curricular planning as its own article, executive functions, artistic creativity, music, motor skills, joint attention, CLASS/serve-and-return interactions, and teacher training—although some appear as category limits. SEL appears only as a lateral outcome of ECEfS (Watt and Frydenberg, 2024), not as an axis.

The search was executed on September 2, 2026 (slot 09:02 America/Mexico_City) on DOI pages, Crossref, Springer, MDPI, Elsevier, Taylor & Francis, JAIR, OECD iLibrary, UNESDOC, NAEYC, and publisher 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 ECEfS / early scientific inquiry / STEM-robotics / SEL as product / green score, and to the caution of not translating eco-tips, footprint scores, or units-by-prompt into pedagogy of care of the environment and nature play. N, d, r, AUC, and DOI were not invented: when an artifact (green-footprint app, computer vision of sustainable acts in the yard) lacks a verified study in the corpus with scoring metrics in ages 3–6, it is discussed as a category ceiling supported by general AI mappings in ECE (Chen, 2024; Su and Yang, 2022; Ljungcrantz, 2026) and by Zhu’s (2025) contrast, not as an invented empirical finding.

4. Case 1. Eco-tips/green-classroom chatbot, green-footprint scoring, computer vision of sustainable acts, or units-by-prompt do not constitute environmental education / ECEfS

Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) saturate the portrait of the artifact ceiling when AI in ECE is presented as if it were environmental education. Chen (2024) maps global affordances of AI in early childhood education: empirical finding of scoping on emerging uses—tutoring, analytics, content generation—not empirical finding that an eco-tips chatbot cultivates place belonging. Su and Yang (2022) review the field of AI in ECE: empirical finding of synthesis on trends, not on nature play. Ljungcrantz (2026) reviews AI–ECE interaction 2020–2024: empirical finding of state of the art, not of situated ECEfS. Pedagogical inference, marked as such: the gesture “green classroom plan from chatbot = environmental education session” is a generative planning ceiling. Ardoin and Bowers (2020) and Ernst et al. (2021) call for nature-rich pedagogies and nature play; a useful chatbot plan can coexist with absence of yard, garden, and adult mediation.

Nikolopoulou (2025) and Miao and Holmes (2023) name the risk of the environmental education units-by-prompt generator. Nikolopoulou (2025) balances promises and challenges of child-centered GenAI: caution framework. Miao and Holmes (2023) demand pedagogical validation and age thresholds for generative AI. Status: normative and review framework, not trial of green units-by-prompt versus nature play. Pedagogical inference: producing a unit by prompt can be subordinate teacher preparation; ECEfS begins when there are hands in soil, water that is cared for, waste that is sorted, and an adult who mediates ethical conversation. Su and Zhong (2022) propose AI curriculum design in ECE as a future direction—AI literacy curriculum, not situated environmental education. Limit inference: an AI curriculum does not sign the garden.

The app that scores a child’s “ecological behavior” or “green footprint” and the computer vision model that classifies “sustainable acts” in the yard lack, in the verified corpus, trials with N, d, r, or AUC reported for sustainability scoring in preschool ages 3–6; figures are not invented. Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) map analytics, sensors, and classification as an AI-in-ECE trend, without equivalence to ECEfS. Zhu (2025) offers the most useful conceptual contrast: AI as Co-Pilot for sustainability education in ECE under the formula empowerment without replacement—teacher support without replacing practice. Restrictive pedagogical inference: teacher co-pilot ≠ eco-tips chatbot directed at the child; empowerment ≠ green score of the infant; without replacement ≠ computer vision that declares sustainable act fulfilled. UNESCO (2021) and U.S. Department of Education (2023) demand human supervision. The four artifacts share the same grammar of substitution: the tip replaces the repeated habit; the score replaces place belonging; the vision label replaces pedagogical observation; the prompt replaces outdoor experience.

It is useful to specify the ceiling without inventing effects. A chatbot can list tips; Ardoin and Bowers (2020) and Lepičnik Vodopivec and Šindić (2025) associate outcomes with situated pedagogies, not lists. An app can sum points; Watt and Frydenberg (2024) measure outcomes in a nature program, not in a ranking. A classifier can label a gesture; Ernst et al. (2021) anchor nature play, not pose detection. A generator can print a unit; Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021) document teacher implementation, not prompt output. Pedagogical inference: “high footprint” or “unit according to the model” is measurement or generation, not signed ECEfS.

5. Case 2. What the early childhood center does when there is environmental education / ECEfS: care of place, nature play, habits and mediation

Ardoin and Bowers (2020) situate ECEE as a reviewable empirical field: synthesis of environmental, cognitive, and socioemotional outcomes under play-based and nature-rich pedagogies. Status: foundational synthesis empirical finding. Pedagogical inference, marked as such: this is the object an early childhood center can call environmental education. ECEfS is not a footprint score: it is growing domain of caring for the nearby environment, exploring nature sensorially, reiterating water, waste, yard, and garden habits, conversing ethically with an adult, and feeling place belonging. An app that marks a “green point” without shared practice can celebrate a labeled gesture and, at the same time, empty the yard of mediation. A sustainable-act classifier can label “recycled” in video and not have captured whether the adult modeled water care, adjusted the group’s ethical language, or invited transferring the habit from the sink to the garden.

Ernst, McAllister, Siklander and Storli (2021) saturate the craft from nature play (32 studies): do not abandon play in soil, leaves, water, and yard for a tip. Lepičnik Vodopivec and Šindić (2025) compare eco-programs versus no eco-program (N = 114 teachers) on ecological, social, and economic dimensions. Watt and Frydenberg (2024) document Learning in Nature (Melbourne; 2021/2022) with pro-environmental outcomes and SEL only as a lateral effect—not the axis of this article. Šindić, Višnjić-Jevtić and Lepičnik Vodopivec (2025) situate children as agents of change; Borg and Pramling Samuelsson (2022) and Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021) anchor agency and teacher implementation; Borg and Gericke (2021) teach social sustainability; Eilam (2023) articulates place-based—TRANSFER marked. NAEYC (2022) and OECD (2021, 2023) anchor exploration, meaningful interactions, and subordinate digitalization.

Mediation—NAEYC (2022), OECD (2021), and Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021)—is adult presence that organizes, models, invites, and amplifies: brings soil, water, or waste close; listens to the ethical question; protects sensory exploration in the yard; reiterates habits in sink, dining room, and garden; documents belonging without ranking. Pedagogical inference: pedagogical observation of care is professional reading, not a log of green points. A dashboard does not model watering; an adult in the garden does. Ernst et al. (2021), Ardoin and Bowers (2020), and Lepičnik Vodopivec and Šindić (2025) orient play ecologically: ECEfS requires nearby environment, body in nature, repeated habit, and adult who mediates—not a tips feed nor footprint score.

6. Case 3. ECEfS / early childhood environmental education is not early scientific inquiry, nor STEM/robotics, nor SEL as a product, nor a green score

The first category boundary is early scientific inquiry. Observing an insect, asking “where does it live?” or recording a finding can coexist with ECEfS; it does not, by itself, constitute environmental education of care of the environment, water/waste habits, and place belonging. Status: pedagogical inference marked as such, anchored in Ardoin and Bowers (2020) and Ernst et al. (2021) as environmental/nature play construct, and in the explicit exclusion of the inquiry axis already published in the series. This article does not convert the science table into central object: it names it to forbid equivalence. An eco-tips chatbot is not “green inquiry” either: it is tip generation. Confusing “we have a green classroom plan” with “there is mediated care practice” is the category error the market exploits.

The second boundary is STEM/robotics. A kit, a school environmental sensor, or a “program the watering” sequence can coexist in a center; they do not, by themselves, constitute ECEfS of nature play and everyday habits at ages 3–6. Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) map AI and technology in ECE; here they are used only as limit: technological affordance ≠ environmental craft. The third boundary is SEL as product. Watt and Frydenberg (2024) show that a Learning in Nature program can associate with SEL outcomes as a lateral effect of ECEfS: empirical finding of empirical overlap, not of construct identity. Restrictive pedagogical inference: this article does not recycle SEL as axis; a dashboard that rewards “positive emotion in the yard” does not convert affective reinforcement into environmental education. The fourth boundary is green score / gamified green footprint. Scoring “ecological behavior” inverts pedagogy: an evaluative proxy is made to stand for situated care. Zhu (2025) rejects, in conceptual framework, replacement of the teacher; UNESCO (2021), Miao and Holmes (2023), and U.S. Department of Education (2023) demand human supervision. Pedagogical inference: the only use of AI coherent with ages 3–6 remains on the side of the adult who organizes yard, garden, and habits—as support for material preparation or deferred professional reflection (Zhu, 2025: Co-Pilot)—, subject to pedagogical validation. It does not enter as autonomous footprint scorer, eco-tips chatbot for the child, classifier of fulfilled sustainable acts, nor generator that replaces outdoor experience.

Eilam (2023)—TRANSFER marked—and Borg and Gericke (2021) reinforce place-based and social sustainability without converting them into score: place belonging is cultivated in ethically mediated conversation and nearby community practice (Šindić, Višnjić-Jevtić and Lepičnik Vodopivec, 2025), not in an app that declares footprint fulfilled.

7. Inferential framework: four tests to claim environmental education / ECEfS, 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 nursery, preschool, CENDI, or infant school does not pass them, it cannot declare that an eco-tips/green-classroom chatbot, a green-footprint app, computer vision of sustainable acts, or an environmental units-by-prompt generator constitute environmental education / ECEfS.

7.1. Test of situated practice of care of the nearby environment, sensory exploration, and everyday habits, not of the eco-tip nor the green classroom plan. Ardoin and Bowers (2020) and Ernst et al. (2021) define the craft as nature-rich pedagogy and nature play. Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) map chatbots and content generation as affordances, not as ECEfS. Pedagogical inference: evidence of environmental education is verified in whether the child cared for the environment, explored sensorially, and reiterated water, waste, yard, or garden habits. If the center’s “evidence” is a printed tip or a generated green classroom plan, the center has done generative planning, not ECEfS.

7.2. Test of the yard, the garden, eco-programs, and place belonging, not of the environmental unit by prompt. Lepičnik Vodopivec and Šindić (2025) situate eco-programs (N = 114 teachers); Watt and Frydenberg (2024) situate Learning in Nature; NAEYC (2022) situates environment exploration. Miao and Holmes (2023) demand pedagogical validation of generative AI. Pedagogical inference: producing a unit by prompt does not demonstrate that there was yard, garden, nature play, or place belonging. A script can exist; it does not sign outdoor practice.

7.3. Test of teacher mediation and ethical conversation, not of the green-footprint dashboard nor the sustainable-act classifier. OECD (2021, 2023) demand meaningful interactions and subordinate digitalization. Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021) and Borg and Pramling Samuelsson (2022) anchor teacher implementation and child agency. Zhu (2025) demands Co-Pilot without replacement. Pedagogical inference: “high green footprint” or “sustainable act detected” can raise a threshold without raising mediation quality. Pedagogical observation reads care and ethical conversation; the dashboard counts a label.

7.4. Test of category distinction and professional judgment, not of the product catalog. ECEfS ≠ early scientific inquiry, STEM/robotics, SEL as product, nor green score. Watt and Frydenberg (2024) prevent confusing lateral SEL with construct identity. Eilam (2023) and Borg and Gericke (2021) prevent reducing place-based or social sustainability to tip. UNESCO (2021), Miao and Holmes (2023), U.S. Department of Education (2023), NAEYC (2022), and OECD (2021, 2023) demand human supervision, pedagogical validation, and not replacing professional judgment. Pedagogical inference: a center cannot treat the infant as a footprint vector nor as emitter of “sustainable acts” for a pipeline. Environmental education is not fulfilled by scoring ecological behavior better. It is fulfilled by practicing care of the nearby environment, sensory exploration, everyday habits, and ethical conversation with adults who mediate in yard, garden, and routines.

The framework admits the digital when subordinated to teacher preparation and validated professional observation (Zhu, 2025, as Co-Pilot; Chen, 2024, Su and Yang, 2022, and Ljungcrantz, 2026, as subordinate affordances). It rejects declaring ECEfS by autonomous eco-tips chatbot, footprint app, computer vision of sustainable acts, or unit generator that replaces outdoor experience (Miao and Holmes, 2023; Nikolopoulou, 2025; UNESCO, 2021).

The four tests are read together. Passing only the first—isolated habit existed—without mediation or category distinction is not enough. Passing only the fourth—the center distinguishes constructs—without situated practice is not enough either.

8. Discussion

Three tensions organize the discussion. The first is between displaying tip, scoring, vision, or generation and exercising ECEfS. It is empirical finding that nature-rich ECEE, nature play (32 studies), eco-programs (N = 114), and Learning in Nature sustain situated outcomes (Ardoin and Bowers, 2020; Ernst et al., 2021; Lepičnik Vodopivec and Šindić, 2025; Watt and Frydenberg, 2024). It is framework that AI in ECE grows without equivalence to situated care (Chen, 2024; Su and Yang, 2022; Ljungcrantz, 2026) and that Co-Pilot must empower without replacing (Zhu, 2025). It is not empirical finding that the four artifacts produce yard craft: they typify, score, label, or generate what engineering delivers and declare what only mediated practice would authorize.

The second is between automated assessment of “ecological behavior” and everyday care pedagogy. Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) objectify analytics and classification affordances as trend; they do not report in the corpus verified trials that validate green-footprint scoring or CV of sustainable acts in preschool ages 3–6—therefore metrics are not invented. Pedagogical inference: insisting that the nursery “already has environmental education” because a model classifies sustainable acts or sums footprint is inverted pedagogy. An algorithmic proxy is made to stand for care practice, sensory exploration, and habits. The infant becomes a signal emitter for the classifier; the adult, a dashboard supervisor.

The third is between generated unit and integral outdoor craft. Miao and Holmes (2023) and Nikolopoulou (2025) set GenAI limits; Su and Zhong (2022) situate AI curriculum as distinct construct. Pedagogical inference: selling an environmental unit by prompt as “ECEfS with AI” confuses textual product with situated practice. An autonomous chatbot because the system needs output does not replace the adult who opens the tap with the child, sorts waste in the yard, and converses about why garden water matters.

Additional pedagogical inference: the artifacts share an economy of visibility—tip, score, label, script—versus ECEfS verified in everyday life of the nearby environment (OECD, 2021; NAEYC, 2022; Ardoin and Bowers, 2020). Watt and Frydenberg (2024) confirm that SEL can be a lateral outcome without being the construct. Borg and Pramling Samuelsson (2022), Šindić, Višnjić-Jevtić and Lepičnik Vodopivec (2025), and Borg and Gericke (2021) confirm situated agency and social sustainability. Miao and Holmes (2023), UNESCO (2021), and U.S. Department of Education (2023) subordinate AI to professional judgment.

The four tests of section 7 read these tensions with operational criterion. 7.1 demands care, sensory exploration, and habits—not eco-tip nor green classroom plan (Ardoin and Bowers, 2020; Ernst et al., 2021). 7.2 asks about yard, garden, eco-program, and place belonging—not only unit by prompt (Lepičnik Vodopivec and Šindić, 2025; Watt and Frydenberg, 2024). 7.3 distinguishes mediation and ethical conversation from footprint dashboard or sustainable-act classifier (OECD, 2021, 2023; Zhu, 2025). 7.4 prevents confusing ECEfS with early scientific inquiry, STEM/robotics, SEL as product, or green score (Watt and Frydenberg, 2024; Eilam, 2023). Pedagogical inference: passing only the fourth test without situated practice displays generative planning, not declared environmental education.

The empirical contrast of craft—nature-rich ECEE (Ardoin and Bowers, 2020); nature play in 32 studies (Ernst et al., 2021); eco-programs with N = 114 teachers (Lepičnik Vodopivec and Šindić, 2025); Learning in Nature (Watt and Frydenberg, 2024); kindergarten–community agency and Swedish EfS (Šindić, Višnjić-Jevtić and Lepičnik Vodopivec, 2025; Borg and Pramling Samuelsson, 2022; Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér, 2021)—defines the floor that the four artifacts do not reach alone. Zhu (2025) situates AI as Co-Pilot without replacement: subordinate teacher support (NAEYC, 2022; OECD, 2021, 2023; Miao and Holmes, 2023), not tip to the child nor green score. UNESCO (2021) and U.S. Department of Education (2023) demand human supervision; Su and Zhong (2022) mark AI literacy as distinct construct. Pedagogical inference: demanding “green classroom with AI” without the four tests inverts the hierarchy: the product commands over the craft.

When a nursery declares environmental education through an eco-tips chatbot, it can replace the water habit with a printed list; when it displays green footprint, it can reward thresholds instead of yard belonging; when it replaces nature play with a generated unit, mud, watering, and ethical conversation disappear (Ernst et al., 2021; Watt and Frydenberg, 2024; Borg and Pramling Samuelsson, 2022). Pedagogical inference: ECEfS erodes not by using AI, but by treating tip, scoring, vision, or generation as if they were the practice. Subordinate digital is legitimate—preparing garden materials with caution—; it is not legitimate to invert the sequence and install an artifact to declare ECEfS fulfilled. It is verified in the yard, with real habits and present adult.

9. Limits

This review is narrative. It does not apply PRISMA nor estimate primary combined effects. Ardoin and Bowers (2020) are foundational systematic review of ECEE, not AI trial. Ernst et al. (2021) review 32 nature play studies, not chatbots nor footprint scoring. Lepičnik Vodopivec and Šindić (2025) survey N = 114 teachers on eco-programs, not chatbot versus mediated yard comparison. Watt and Frydenberg (2024) report Learning in Nature (2021/2022 cohorts) with lateral SEL: not an article of SEL as axis. Šindić, Višnjić-Jevtić and Lepičnik Vodopivec (2025) address kindergarten–community agency. Borg and Pramling Samuelsson (2022), Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér (2021), and Borg and Gericke (2021) are from Swedish EfS context, without AI as primary variable. Eilam (2023) is place-based with TRANSFER marked due to mixed age. Chen (2024), Su and Yang (2022), Su and Zhong (2022), Ljungcrantz (2026), and Nikolopoulou (2025) map AI in ECE, not ECEfS as primary variable with footprint scoring. Zhu (2025) is conceptual Co-Pilot framework, not controlled trial. No verified trials of green-footprint app or computer vision scoring sustainable acts in CENDI ages 3–6 were located in the corpus; those artifacts are discussed as category ceiling, not as invented finding. NAEYC, UNESCO, and OECD are framework sources. No Latin American trials comparing adult-mediated care versus autonomous eco-tips chatbot in nursery were located. Section 7 inferences are pedagogical category hypotheses, not implementation evidence.

10. Conclusions

A chatbot that gives “eco-tips” or “green classroom” plans, an app that scores a child’s “ecological behavior” or “green footprint,” a computer vision model that classifies “sustainable acts” in the yard, or a generator that produces environmental education units from a prompt do not constitute environmental education nor education for sustainability in an early childhood center. Verified evidence does not authorize that declaration. Chen (2024), Su and Yang (2022), and Ljungcrantz (2026) confirm AI affordances in ECE without equivalence to situated care. Nikolopoulou (2025) and Miao and Holmes (2023) set GenAI limits. Zhu (2025) demands Co-Pilot without replacement—contrast: teacher support ≠ tip to the child nor green score. Conversely, when there is ECEfS in early childhood, there is situated practice: nature-rich ECEE (Ardoin and Bowers, 2020); nature play (Ernst et al., 2021); eco-programs N = 114 (Lepičnik Vodopivec and Šindić, 2025); Learning in Nature with lateral SEL (Watt and Frydenberg, 2024); agency and teacher implementation (Borg and Pramling Samuelsson, 2022; Šindić et al., 2025; Engdahl, Pramling Samuelsson and Ärlemalm-Hagsér, 2021; Borg and Gericke, 2021); place-based—TRANSFER marked—(Eilam, 2023); DAP and interactions (NAEYC, 2022; OECD, 2021, 2023). ECEfS is distinguished from early scientific inquiry, STEM/robotics, SEL as product, and green score. Guidelines demand human supervision and not replacing professional judgment (UNESCO, 2021; Miao and Holmes, 2023; U.S. Department of Education, 2023; Su and Zhong, 2022).

Where sources do not measure a nursery, this article does not claim it. Where they measure AI mappings, GenAI, or Co-Pilot frameworks, it does not translate them into pedagogical environmental education. Accompanying three- to six-year-olds in sustainability is exercising care of the nearby environment, sensory exploration of nature, everyday habits of water, waste, yard, and garden, ethical conversation, and place belonging with adult mediation. The rest is generated eco-tip, green classroom plan by prompt, footprint scoring, and algorithmic classification of sustainable acts. It is not environmental education / ECEfS in early childhood, and it must not be presented as what it is not.

Laboratorio Editorial de NEXTECH.IA / Ingeniero Mitre.

References

  1. Ardoin, N. M., y Bowers, A. W. (2020). Early childhood environmental education: A systematic review of the research literature. Educational Research Review, 31, 100353. https://doi.org/10.1016/j.edurev.2020.100353
  2. Borg, F., y Gericke, N. (2021). Local and global aspects: Teaching social sustainability in Swedish preschools. Sustainability, 13(7), 3838. https://doi.org/10.3390/su13073838
  3. Borg, F., y Pramling Samuelsson, I. (2022). Preschool children’s agency in education for sustainability: the case of Sweden. European Early Childhood Education Research Journal, 30(1), 147–163. https://doi.org/10.1080/1350293X.2022.2026439
  4. Chen, J. J. (2024). A scoping study on AI affordances in early childhood education. Journal of Artificial Intelligence Research, 81, 701–740. https://doi.org/10.1613/jair.1.16882
  5. Eilam, E. (2023). Exploring the affordances of place-based education for advancing sustainability education: The role of cognitive, socio-emotional and behavioural learning. Education Sciences, 13(7), 676. https://doi.org/10.3390/educsci13070676
  6. Engdahl, I., Pramling Samuelsson, I., y Ärlemalm-Hagsér, E. (2021). Swedish teachers in the process of implementing education for sustainability in early childhood education. New Ideas in Child and Educational Psychology, 1(1), 3–23. https://doi.org/10.11621/nicep.2021.0101
  7. Ernst, J., McAllister, K., Siklander, P., y Storli, R. (2021). Contributions to sustainability through young children’s nature play: A systematic review. Sustainability, 13(13), 7443. https://doi.org/10.3390/su13137443
  8. Lepičnik Vodopivec, J., y Šindić, A. (2025). Early childhood education for sustainability in preschools with eco-programs: Implementation and learning outcomes. Journal of Baltic Science Education, 24(1), 133–148. https://doi.org/10.33225/jbse/25.24.133
  9. 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
  10. Miao, F., y Holmes, W. (2023). Guidance for generative AI in education and research. UNESCO. https://doi.org/10.54675/EWZM9535
  11. 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
  12. 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
  13. OECD. (2021). Starting Strong VI: Supporting meaningful interactions in early childhood education and care. OECD Publishing. https://doi.org/10.1787/f47a06ae-en
  14. OECD. (2023). Empowering young children in the digital age (Starting Strong). OECD Publishing. https://doi.org/10.1787/50967622-en
  15. Šindić, A., Višnjić-Jevtić, A., y Lepičnik Vodopivec, J. (2025). ECEfS from kindergarten to community: Children as agents of change in education for sustainable development. International Journal of Early Childhood, 57, 939–957. https://doi.org/10.1007/s13158-025-00432-6
  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. UNESCO. (2021). Recommendation on the ethics of artificial intelligence. UNESCO. https://unesdoc.unesco.org/ark:/48223/pf0000381137
  20. Watt, B. A., y Frydenberg, E. (2024). Early childhood education for sustainability: Outcomes for social and emotional learning. Australasian Journal of Early Childhood, 50(2), 131–145. https://doi.org/10.1177/18369391241287939
  21. Zhu, H. (2025). AI as a Co-Pilot: Framework for sustainability education in early childhood. Advances in Management and Intelligent Technologies, 1(7). https://doi.org/10.62177/amit.v1i7.1114