Future of Education with Neuro-Symbolic AI Agents in Self-Improving Adaptive Instructional Systems

Abstract

This paper proposes a novel approach to use artificial intelligence (AI), particularly large language models (LLMs) and other foundation models (FMs) in an educational environment. It emphasizes the integration of teams of teachable and self-learning LLMs agents that use neuro-symbolic cognitive architecture (NSCA) to provide dynamic personalized support to learners and educators within self-improving adaptive instructional systems (SIAIS). These systems host these agents and support dynamic sessions of engagement workflow. We have developed the never ending open learning adaptive framework (NEOLAF), an LLM-based neuro-symbolic architecture for self-learning AI agents, and the open learning adaptive framework (OLAF), the underlying platform to host the agents, manage agent sessions, and support agent workflows and integration. The NEOLAF and OLAF serve as concrete examples to illustrate the advanced AI implementation approach. We also discuss our proof of concept testing of the NEOLAF agent to develop math problem-solving capabilities and the evaluation test for deployed interactive agent in the learning environment.

Graphical abstract

Keywords

large language models (LLMs) / neuro-symbolic cognitive architecture (NSCA) / adaptive instructional systems (AIS) / open learning adaptive framework (OLAF) / never ending open learning adaptive framework (NEOLAF) / artificial intelligence in education (AIED) / intelligent tutoring system (ITS) / LLM agent

References

[1] Abd-Alrazaq A., AlSaad, R., Alhuwail, D., Ahmed, A., Healy, P. M., Latifi, S., Aziz, S., Damseh, R., Alrazak, S. A., &Sheikh J. (2023). Large language models in medical education: Opportunities, challenges, and future directions.JMIR Medical Education, 9: e48291

[2] AhoA. V., Lam, M. S., Sethi, R., &UllmanJ. D.. (2006). Compilers: Principles, techniques, & tools. Boston: Addison-Wesley.

[3] BenderE. M., Gebru, T., McMillian-Major, A., &ShmitchellS.. (2021). On the dangers of stochastic parrots: Can language models be too big? In: Proceedings of 2021 ACM Conference on Fairness, Accountability, and Transparency. ACM, 610–623.

[4] Bengio Y., Ducharme, R., Vincent, P., &Janvin C. (2003). A neural probabilistic language model.The Journal of Machine Learning Research, 3: 1137–1155

[5] BoochG., Fabiano, F., Horesh, L., Kate, K., Lenchner, J., Linck, N., Loreggia, A., Murugesan, K., Mattei, N., Rossi, F., &SrivastavaB.. (2020). Thinking fast and slow in AI. arXiv Preprint, arXiv:2010.06002.

[6] Brown T. B., Mann, B., Ryder, N., Subbiah, M., Kaplan, J., Dhariwal, P., Neelakantan, A., Shyam, P., Sastry, G., Askell, A., Agarwal, S., Herbert-Voss, A., Krueger, G., Henighan, T., Child, R., Ramesh, A., Ziegler, D. M., Wu, J., Winter, C., Hesse, C., Chen, M., Sigler, E., Litwin, M., Gray, S., Chess, B., Clark, J., Berner, C., McCandlish, S., Radford, A., Sutskever, I., &Amodel D. (2020). Language models are few-shot learners.Advances in Neural Information Processing Systems, 33: 1877–1901

[7] DevlinJ., Chang, M., Lee, K., &ToutanovaK.. (2018). BERT: Pre-training of deep bidirectional transformers for language understanding. arXiv Preprint, arXiv:1810.04805.

[8] JoublinF., Ceravola, A., Deigmoeller, J., Gienger, M., Franzius, M., &EggertJ.. (2023). A glimpse in ChatGPT capabilities and its impact for AI research. arXiv Preprint, arXiv:2305.06087.

[9] JurafskyD., Martin, J. H., Kehler, A., Linden, K. V., &WardN.. (2009). Speech and language processing: An introduction to natural language processing, computational linguistics, and speech recognition. New Jersey: Prentice Hall.

[10] KahnemanD.. (2011). Thinking, fast and slow. New York: Farrar, Straus and Giroux.

[11] LeikerD., Finnigan, S., Gyllen, A. R., &CukurovaM.. (2023). Prototyping the use of large language models (LLMs) for adult learning content creation at scale. arXiv PrePrint, arXiv:2306.01815.

[12] LuitseD., &DenkenaW.. (2021). The great transformer: Examining the role of large language models in the political economy of AI. Big Data & Society, 8(2).

[13] MikolovT., Chen, K., Corrado, G., &DeanJ.. (2013). Efficient estimation of word representations in vector space. arXiv Preprint, arXiv:1301.3781.

[14] Mitchell T., Cohen, W., Hruschka, E., Talukdar, P., Yang, B., Betteridge, J., Carlson, A., Dalvi, B., Gardner, M., Kisiel, B., Krishnamurthy, J., Lao, N., Mazaitis, K., Mohamed, T., Nakashole, N., Platanios, E., Ritter, A., Samadi, M., Settles, B., Wang, R., Wijaya, D., Gupta, A., Chen, X., Saparov, A., Greaves, M., &Welling J. (2018). Never-ending learning.Communication of the ACM, 61(5): 103–115

[15] RadfordA., Wu, J., Child, R., Luan, D., Amodei, D., &SutskeverI.. (2019). Language models are unsupervised multitask learners. OpenAI.

[16] SchramowskiP., Turan, C., Andersen, N., Rothkopf, C., &KerstingK.. (2021). Language models have a moral dimension. arXiv Preprint, arXiv:2103.11790.

[17] Tong R. J., &Lee T. X. (2023). Trustworthy AI that engages humans as partners in teaching and learning.Computer, 56(5): 62–73

[18] TuX., Zou, J., Su, W., &ZhangL.. (2023). What should data science education do with large language models? arXiv Preprint, arXiv:2307.02792.

[19] VaswaniA., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.. N., Kaiser, L., & Polosukhin, I. (2017). Attention is all you need. arXiv Preprint, arXiv:1706.03762.

[20] WeiJ., Bosma, M., Zhao, V. Y., Guu, K., Yu, A. W., Lester, B., Du, N., Dai, A. M., Le, Q. V.. (2021). Finetuned language models are zero-shot learners. arXiv Preprint, arXiv:2109.01652.

[21] YadavG.. (2023). Scaling evidence-based instructional design expertise through large language models. arXiv Preprint, arXiv:2306.01006.

[22] Yan L., Sha, L., Zhao, L., Li, Y., Martínez-Maldonado, R., Chen, G., Li, X., Jin, Y., &Gašević D. (2023). Practical and ethical challenges of large language models in education: A systematic scoping review.British Journal of Educational Technology, 55: 90–112

[23] ZhaoW. X., Zhou, K., Li, J., Tang, T., Wang, X., Hou, Y., Min, Y., Zhang, B., Zhang, J., Dong, Z., Du, Y., Yang, C., Chen, Y., Jiang, J., Ren, R., Li, Y., Tang, X., Liu, Z., Liu, P., Nie, J., & Wen, J. (2023). A survey of large language models. arXiv Preprint, arXiv:2303.18223.

Acknowledgments

We thank Squirrel AI for providing the deployment and testing environment for the interactive agents and their collaboration support. The process of writing this paper was greatly aided by generative AI models, including (but not exclusively) Microsoft’s Bing, Anthropic’s Claude, Mistral AI’s Mistral, Google’s Bard, Meta’s LLaMA, and OpenAI’ s GPT models.

Conflict of Interest

The authors declare that they have no conflict of interest.

RIGHTS & PERMISSIONS

2024 Higher Education Press