Findings — SELBI

Phase 1 · Baseline Survey

The Blue Skills Knowledge Gap

A nationwide survey of 967 ECE educators (n=137, 95% CI) across KG1, KG2, Year 1 and Year 2 revealed a significant knowledge gap and strong demand for support.

Blue Skills Familiarity (1–5 scale)

1 — Not familiar

61%

2

11%

3

17%

4

9%

5 — Very familiar

2%

Mean familiarity score: 1.80 / 5.00

Respondent Profile

State schools

69%

Church schools

23%

Private schools

8%

Year Group Distribution

KG1

32%

KG2

34%

Year 1

18%

Year 2

16%

Educator Attitudes

What Educators Told Us

Percentage of respondents who agreed or strongly agreed with key statements.

82%

want access to training on integrating marine sustainability into lessons

76%

report lacking resources such as teaching guides, materials, or digital tools

82%

agree Malta's sea access makes early blue skills education essential

Phase 2 & 3 · Internal Evaluation

Platform Evaluation Results

Two-phase expert evaluation (TRL 4) using an 11-criterion rubric. Phase 1 tested 16 experiments; Phase 2 introduced 20 more complex, scenario-based prompts.

Absence of Hallucinations

5.00 / 5

Perfect score in Phase 2 — zero fabricated content across all 29 ratings

Content Correctness

4.96 / 5

Scientific accuracy confirmed by the blue skills expert

Pedagogical Appropriateness

4.85 / 5

Age-appropriate and aligned with ECE practice (Phase 1)

Engagement Factor

5.00 / 5

Perfect engagement scores — outputs were pedagogically compelling

Content Accuracy & Relevance

4.71 / 5

Highest-scoring dimension overall across both phases

Grand Mean (Phase 1)

4.74 / 5

78.3% of all individual ratings at maximum score of 5

"A highly competent 3-session sequence that balances curiosity, creativity and environmental awareness."

— Dr Heathcliff Schembri, ECE Expert

Final Evaluation

Classroom Deployment Insights

Key findings from the final project evaluation with participating ECE educators across four year groups.

Criterion & Content Validity

Participating teachers were receptive to the technology. The platform generated new ideas beyond common knowledge and adapted the same theme for different year levels and varying skill levels within each year.

SELBI-generated content was effectively adapted across subjects — integrating mathematics, English, and social studies. The platform also successfully adapted activities for students with specific needs, such as a student who could not touch sand during a beach-themed activity.

The system prompts went through ten iterations. The final version strictly limited scope to the curated repository, confirmed by expert evaluators who gave a perfect score for absence of hallucinations.

"We were never aware of how to integrate blue in the teaching content as was demonstrated by SELBI."

— Participating ECE Educator

Face & Construct Validity

Teachers expressed positive perceptions of SELBI as useful and engaging, while suggesting enhancements such as increased interactivity and differentiated modes (teacher, parent, child).

Teachers found the platform reliable precisely because it was not connected to the internet — reducing the risk of misinformation often encountered with other AI tools.

Educators described how SELBI supported children's understanding through learning invitations, play-based engagement, and cross-curricular experiences.