arxiv: 2604.01396 · v2 · submitted 2026-04-01 · 💻 cs.CY · cs.ET· cs.HC

Recognition: no theorem link

Democratizing Foundations of Problem-Solving with AI: A Breadth-First Search Curriculum for Middle School Students

Griffin Pitts , Kimia Fazeli , Tirth Bhatt , Jennifer Albert , Marnie Hill , Tiffany Barnes , Shiyan Jiang , Bita Akram

Authors on Pith no claims yet

Pith reviewed 2026-05-13 21:45 UTC · model grok-4.3

classification 💻 cs.CY cs.ETcs.HC

keywords AI educationK-12 curriculumBreadth-First Searchmiddle school scienceproblem-solvingcontact tracingunplugged activitiesvirus spread

0 comments

The pith

A middle school science curriculum uses Breadth-First Search to teach AI problem-solving through virus spread and contact tracing.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper presents a curriculum that embeds AI learning goals into existing middle school science classes by introducing Breadth-First Search as a way to explore networks and find shortest paths. Students complete unplugged activities and use an interactive simulation to apply the method to real science topics such as how viruses move through contacts. Pre- and post-assessments along with student work and teacher feedback indicate productive engagement and gains in understanding both BFS and AI problem-solving. The approach shows that these concepts can be learned productively without separate AI classes or coding requirements. Teacher observations confirm the module fits within ongoing science instruction and supports science outcomes at the same time.

Core claim

By designing unplugged activities and a simulation around Breadth-First Search, the curriculum lets middle school students learn AI problem-solving strategies while studying virus spread and contact tracing in their regular science class, producing measurable gains in BFS understanding and positive teacher reports on integration fit.

What carries the argument

The integrated BFS module of unplugged network exploration activities plus an interactive simulation that students apply directly to science models of virus transmission.

If this is right

Students gain the ability to apply BFS to identify shortest paths in networks modeling real-world spread.
AI problem-solving foundations become accessible inside standard subject classes without dedicated AI time or programming.
Science learning outcomes remain supported while AI concepts are introduced.
Rural and under-resourced classrooms can add AI exposure using only existing science units and low-tech activities.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Similar search-algorithm modules could be adapted for other middle-school subjects such as history or geography to reach more students.
The unplugged-plus-simulation design lowers barriers for teachers who lack AI background.
Wider use might shift K-12 AI education from elective add-ons toward routine integration in core subjects.
Longer-term tracking could test whether early BFS exposure improves later performance on more advanced AI topics.

Load-bearing premise

Improvements seen on pre- and post-assessments come from the curriculum itself rather than from practice effects of taking the test or other unmeasured classroom factors.

What would settle it

A comparison study in which students who receive the BFS module show no greater gains on BFS and AI questions than a matched group that receives only standard science instruction on the same topics would falsify the learning claim.

Figures

Figures reproduced from arXiv: 2604.01396 by Bita Akram, Griffin Pitts, Jennifer Albert, Kimia Fazeli, Marnie Hill, Shiyan Jiang, Tiffany Barnes, Tirth Bhatt.

**Figure 1.** Figure 1: Interface views from the I-SAIL BFS activity sequence information from what is already known [16]. The module was implemented as a seven-day instructional sequence totaling approximately five hours. It began with an introduction to AI and BFS-related concepts through lecture and guided discussion. Students then completed four simulation-based activities in i-SAIL (Section 3.1), and the remaining sessions e… view at source ↗

read the original abstract

As AI becomes more common in students' everyday experiences, a major challenge for K-12 AI education is designing learning experiences that can be meaningfully integrated into existing subject-area instruction. This paper presents the design and implementation of an AI4K12-aligned curriculum that embeds AI learning goals within a rural middle school science classroom using Breadth-First Search (BFS) as an accessible entry point to AI problem-solving. Through unplugged activities and an interactive simulation environment, students learned BFS as a strategy for exploring networks and identifying shortest paths, then applied it to science contexts involving virus spread and contact tracing. To examine engagement and learning, we analyzed pre- and post-assessments, student work artifacts, and a teacher interview. Results suggest that students engaged productively with the curriculum, improved their understanding of BFS and AI problem-solving, and benefited from learning these ideas within ongoing science instruction. Teacher feedback further indicated that the module fit well within the science curriculum while supporting intended science learning outcomes. We conclude with curriculum and design considerations for broadening access to learning about problem-solving with AI in education.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A practical curriculum example for BFS in middle school science, but the pre/post gains lack controls and details to support the learning claims.

read the letter

This paper gives a specific classroom implementation of a BFS module for middle school science, using unplugged activities and a simulation tied to virus spread and contact tracing. It aligns with the AI4K12 framework and shows how to embed the topic in existing rural instruction without adding separate AI time. That integration is the main practical contribution and addresses a real constraint for many schools. The design choices, like starting with networks and shortest paths before the science application, are straightforward and age-appropriate. The teacher interview notes a good fit with science goals, which is a positive sign for feasibility. The evaluation is the soft spot. Pre- and post-assessments plus artifacts indicate productive engagement and some understanding gains, but the report gives no sample size, no statistical tests, and no control condition. As the stress-test note points out, this leaves practice effects, prior knowledge, or regular class time as plausible alternative explanations for any improvement. The evidence is descriptive rather than comparative. This is for teachers and curriculum developers who want a ready template for early AI problem-solving in science classes. It is not positioned as a rigorous outcome study, so readers should treat the gains as preliminary indicators. I would bring it to a reading group on K-12 AI education to discuss the activity sequence and how it might scale. It deserves peer review because the curriculum details are concrete enough to be useful to others, and referees can ask for clearer reporting on the evaluation or limitations.

Referee Report

1 major / 0 minor

Summary. The manuscript presents the design and implementation of an AI4K12-aligned curriculum that introduces Breadth-First Search (BFS) as an entry point to AI problem-solving for middle school students. It embeds unplugged activities and an interactive simulation within a rural science classroom, connecting BFS to network exploration, shortest paths, virus spread, and contact tracing. Evaluation draws on pre- and post-assessments, student work artifacts, and a teacher interview, reporting productive engagement, gains in BFS and AI understanding, and successful integration with ongoing science instruction.

Significance. If the reported gains can be robustly attributed to the curriculum, the work offers a practical model for integrating foundational AI algorithms into existing K-12 subject-area teaching, especially in under-resourced rural contexts. It provides concrete design considerations for broadening access to AI problem-solving concepts while aligning with science learning goals.

major comments (1)

[Evaluation section] Evaluation section: The central claim that students 'improved their understanding of BFS and AI problem-solving' rests on pre- and post-assessments and artifacts, yet the manuscript reports no sample size, statistical tests, control condition, or validation details for the assessments. This leaves the attribution of gains vulnerable to practice effects, prior knowledge, or unmeasured classroom variables, directly undermining the strength of the empirical conclusions.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive comments on the evaluation. We will revise the manuscript to increase transparency around methods, sample details, and limitations while preserving the exploratory character of the work.

read point-by-point responses

Referee: [Evaluation section] Evaluation section: The central claim that students 'improved their understanding of BFS and AI problem-solving' rests on pre- and post-assessments and artifacts, yet the manuscript reports no sample size, statistical tests, control condition, or validation details for the assessments. This leaves the attribution of gains vulnerable to practice effects, prior knowledge, or unmeasured classroom variables, directly undermining the strength of the empirical conclusions.

Authors: We agree that the current reporting is insufficient. The study was conducted in a single rural middle-school science class (N=22 students). Pre- and post-assessments were short open-ended and multiple-choice items developed in consultation with the teacher; we will add sample items, rubrics, and a brief description of how they were piloted. No formal statistical tests were run because of the small sample and the design-based, qualitative emphasis of the research. A control condition was not feasible without removing students from their regular science instruction. In the revision we will (1) state the sample size explicitly, (2) include a dedicated limitations subsection that discusses practice effects, prior knowledge, and classroom confounds, and (3) temper all claims to indicate that the data suggest productive engagement and learning gains supported by artifacts and the teacher interview, rather than asserting robust causal improvement. revision: partial

Circularity Check

0 steps flagged

Empirical curriculum evaluation with no derivations or self-referential reductions

full rationale

The paper describes curriculum design, unplugged activities, a simulation, and reports outcomes from pre/post assessments, student artifacts, and a teacher interview. No equations, parameters, or mathematical derivations appear in the text. Claims rest on descriptive analysis of observed data rather than any reduction of a 'prediction' to fitted inputs or self-citation chains. The central attribution of learning gains is an empirical claim open to methodological critique but does not exhibit circularity by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard educational research assumptions about the validity of pre/post assessments and the benefits of integrating computational thinking into science instruction; no free parameters or new entities are introduced.

axioms (2)

domain assumption Pre- and post-assessments validly measure student understanding of BFS and AI problem-solving
Reported improvements in understanding depend directly on these assessments accurately capturing learning gains.
domain assumption Integrating AI concepts into ongoing science instruction supports both AI and science learning outcomes
The curriculum design and positive teacher feedback rely on this cross-disciplinary integration being effective.

pith-pipeline@v0.9.0 · 5518 in / 1399 out tokens · 49190 ms · 2026-05-13T21:45:07.999362+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

22 extracted references · 22 canonical work pages

[1]

In: Proceedings of the AAAI Conference on Artificial Intelligence

Akram, B., Yoder, S., Tatar, C., Boorugu, S., Aderemi, I., Jiang, S.: Towards an ai- infused interdisciplinary curriculum for middle-grade classrooms. In: Proceedings of the AAAI Conference on Artificial Intelligence. vol. 36, pp. 12681–12688 (2022)

work page 2022
[2]

Middle Grades Review4(3), n3 (2018)

Boulden, D.C., Wiebe, E., et al.: Computational thinking integration into middle grades science classrooms: Strategies for meeting the challenges. Middle Grades Review4(3), n3 (2018)

work page 2018
[3]

ASCD (2021)

Fisher, D., Frey, N.: Better learning through structured teaching: A framework for the gradual release of responsibility. ASCD (2021)

work page 2021
[4]

The Science Teacher91(2), 35–42 (2024)

Hollands, F., Breazeal, C.: Establishing ai literacy before adopting ai. The Science Teacher91(2), 35–42 (2024)

work page 2024
[5]

In: Proceedings of the 16th International Conference of the Learning Sciences-ICLS 2022, pp

Jiang, S., DesPortes, K., et al.: Agents, models, and ethics: Importance of inter- disciplinary explorations in ai education. In: Proceedings of the 16th International Conference of the Learning Sciences-ICLS 2022, pp. 1763-1770. International So- ciety of the Learning Sciences (2022)

work page 2022
[6]

In: Frontiers in Education

Kim, J., Wargo, E.: Empowering educational leaders for ai integration in rural stem education: challenges and strategies. In: Frontiers in Education. vol. 10 (2025)

work page 2025
[7]

In: proceedings of the AAAI Conference on Artificial Intelligence

Kim, S., Jang, Y., Kim, W., et al.: Why and what to teach: Ai curriculum for ele- mentary school. In: proceedings of the AAAI Conference on Artificial Intelligence. vol. 35, pp. 15569–15576 (2021)

work page 2021
[8]

Acm Inroads2(1), 32–37 (2011)

Lee, I., Martin, F., Denner, J., et al.: Computational thinking for youth in practice. Acm Inroads2(1), 32–37 (2011)

work page 2011
[9]

Computers and Education: Artificial Intelligence6, 100217 (2024)

Li, L., Yu, F., Zhang, E.: A systematic review of learning task design for k-12 ai education: Trends, challenges, and opportunities. Computers and Education: Artificial Intelligence6, 100217 (2024)

work page 2024
[10]

In: Proceedings of the 19th WiPSCE Conference on Primary and Secondary Comput- ing Education Research

Morales-Navarro, L., Kafai, Y.B.: Unpacking approaches to learning and teaching machine learning in k-12 education: Transparency, ethics, and design activities. In: Proceedings of the 19th WiPSCE Conference on Primary and Secondary Comput- ing Education Research. pp. 1–10 (2024)

work page 2024
[11]

In: 2025 ASEE Annual Conference & Exposition (2025)

Pitts, G., Marcus, V.M., Motamedi, S.: Student perspectives on the benefits and risks of ai in education. In: 2025 ASEE Annual Conference & Exposition (2025)

work page 2025
[12]

arXiv preprint arXiv:2602.20547 , year=

Pitts, G., Motamedi, S.: What drives students’ use of ai chatbots? technology acceptance in conversational ai. arXiv preprint arXiv:2602.20547 (2026)

work page arXiv 2026
[13]

Rao, A., Piryani, K., et al.: Leveraging large language models to promote ai-infused stem problem-solving for middle school students (2025)

work page 2025
[14]

In: Proceedings of mensch und computer, pp

Scheidt, A., Pulver, T.: Any-cubes: A children’s toy for learning ai: Enhanced play with deep learning and mqtt. In: Proceedings of mensch und computer, pp. 893–895 (2019)

work page 2019
[15]

Computers and Education: Artificial Intelligence 3, 100072 (2022)

Su, J., Zhong, Y.: Artificial intelligence (ai) in early childhood education: Curricu- lum design and future directions. Computers and Education: Artificial Intelligence 3, 100072 (2022)

work page 2022
[16]

Touretzky, D., Gardner-McCune, C., Martin, F., Seehorn, D.: Envisioning ai for k- 12:Whatshouldeverychildknowaboutai?In:ProceedingsoftheAAAIconference on artificial intelligence. vol. 33, pp. 9795–9799 (2019)

work page 2019
[17]

In: 2017 IEEE blocks and beyond workshop (B&B)

Touretzky, D.S.: Computational thinking and mental models: From kodu to ca- lypso. In: 2017 IEEE blocks and beyond workshop (B&B). pp. 71–78. IEEE (2017)

work page 2017
[18]

Chemistry Education Research and Practice13(4), 410–419 (2012) 10 G

Vaino, K., Holbrook, J., Rannikmäe, M.: Stimulating students’ intrinsic motiva- tion for learning chemistry through the use of context-based learning modules. Chemistry Education Research and Practice13(4), 410–419 (2012) 10 G. Pitts et al

work page 2012
[19]

International Journal of Artificial Intelligence in Education33(2), 325–383 (2023)

Williams, R., Ali, S., Devasia, N., et al.: Ai+ ethics curricula for middle school youth: Lessons learned from three project-based curricula. International Journal of Artificial Intelligence in Education33(2), 325–383 (2023)

work page 2023
[20]

In: Proceedings of the AAAI conference on artificial intelligence

Williams, R., Park, H.W., Oh, L., Breazeal, C.: Popbots: Designing an artificial intelligence curriculum for early childhood education. In: Proceedings of the AAAI conference on artificial intelligence. vol. 33, pp. 9729–9736 (2019)

work page 2019
[21]

Yang, W.: Artificial intelligence education for young children: Why, what, and how in curriculum design and implementation. Comput. Educ. Artif. Intell.3(2022)

work page 2022
[22]

In: 5th Workshop on computer science educational data mining (2020)

Yoder, S., Tatar, C., Aderemi, I., Boorugu, S., Jiang, S., Akram, B.: Gaining insight into effective teaching of ai problem-solving through csedm: A case study. In: 5th Workshop on computer science educational data mining (2020)

work page 2020