Artificial intelligence as a double-edged sword: Wielding the POWER principles to maximize its positive effects and minimize its negative effects

Abstract

Advances in artificial intelligence (AI) over the last few decades are transforming the world, pervading nearly all sectors of society, including education, and many aspects of life. In the education discourse, interest in artificial intelligence has sparked various reactions and controversies—everything from appreciation for AI's capabilities to make teaching and student learning more efficient and effective to apprehension about their potential overuse and misuse. In this article, the authors discuss how artificial intelligence is a double-edged sword in early childhood education by presenting some of its positive effects (personalized learning, personalized interactive support, and increased accessibility to broadened learning experiences) and negative effects (overuse and misuse). Considering that young children are growing up in a nearly AI-ubiquitous world and are likely exposed to AI-powered tools, the authors propose applying the POWER (purposeful, optimal, wise, ethical, responsible) principles to maximize the benefits and minimize the drawbacks of AI use. Additionally, the authors recommend the integration of the POWER principles into AI literacy as an imperative for promoting the appropriate use of AI-powered tools.

Keywords

AI literacy artificial intelligence (AI)early childhood education POWER (purposeful optimal wise ethical responsible) principles young children

Artificial intelligence (AI) may be referred to broadly and simply as a technology that imitates and simulates human intelligence to perform complex tasks that would typically require the cognitive capabilities of human beings (e.g. Abbass, 2021; Calo, 2017; McCarthy, 2007). AI is a term that is believed to have been first coined by McCarthy et al. (1955) in their written proposal for the famous summer research conference on AI held in 1956 at Dartmouth College in the United States. However, the concepts underlying AI can be traced back nearly a decade earlier to the seminal work of Alan Turing, a noted British computer scientist. In 1947, Turing delivered a lecture where he publicly mentioned computer intelligence and, in 1948, he wrote a technical report titled “Intelligent machinery” (albeit unpublished), which is considered “the first manifesto of AI” (Copeland, 2004: 2). Since then, Turing’s pioneering conceptualization of AI has paved the way for others to subsequently reinvent the field into a sub-discipline of computer science. This development has also led to the invention of various technologies associated with AI for practical application, thereby demonstrating a proof of concept.

In recent years, we have witnessed burgeoning practical breakthroughs in the invention and deployment of emerging AI-powered technologies for various purposes and in various contexts. For example, at the time of writing, the latest innovation is the Chat Generative Pre-trained Transformer (ChatGPT, https://chat.openai.com), launched in November 2022 by OpenAI (https://openai.com/blog/chatgpt). ChatGPT is essentially an AI language model trained to act as a chatbot that fulfills a myriad of functions, such as generating answers to questions and prompts, providing translations, and adapting to learning based on user feedback. Since its debut, ChatGPT has been widely popular in its usefulness, but its ethical use remains wildly controversial in the public discourse. ChatGPT is just a recent example suggesting that AI is a double-edged sword—a phenomenon that is especially evident in the educational use of AI-generative tools. To contribute to this critical and timely education discourse, we discuss a few of the most noticeable positive and negative effects of AI use. We then present the five interconnected POWER (purposeful, optimal, wise, ethical, responsible) core principles for conceptualizing, guiding, and analyzing the appropriate use of AI to maximize its positive effects and minimize its negative effects.

AI as a double-edged sword

Young children are growing up in a world where the presence of AI is nearly ubiquitous. Notably, AI-powered tools are becoming increasingly popular in early childhood education, as they offer innovative and effective ways to engage young children, support their learning and development, and enhance the quality of their educational experience (e.g. Kewalramani et al., 2021; Lin et al., 2020; Prentzas, 2013; Vartiainen et al., 2020; Williams et al., 2019). The possibilities abound for the use of AI to benefit children's early education. Yet there are also drawbacks, giving rise to ethical concerns and considerations (Adams et al., 2022; Jobin et al., 2019). While this article is not meant to provide an exhaustive list of all the positive and negative effects, we offer a few examples worthy of consideration.

The positive effects of AI in early childhood education

Early childhood education refers to the educational programs and experiences designed to serve young children, such as those aged 0-8 in the United States (NAEYC, 2022). Teachers play a vital role in scaffolding these children's learning within their zone of proximal development, which is defined as the space between what a child can accomplish independently and what a child can achieve with assistance from more competent individuals (e.g. teachers, parents, or peers; Vygotsky, 1978). The emerging AI-powered adaptive learning technologies can serve as scaffolds to enhance a child's learning within one's zone of proximal development (Vygotsky, 1978; Wertsch, 1984). Prior research (e.g. Kewalramani et al., 2021; Lin et al., 2020; Prentzas, 2013; Vartiainen et al., 2020; Williams et al., 2019) has identified positive effects of AI on children's learning and development by serving in scaffolding roles (e.g. a tutor/assistant or humanlike conversational agent) to provide personalized or individualized learning, personalized interactive support, and increased accessibility to broadened learning experiences.

Personalized learning

Adaptive learning platforms use AI algorithms to analyze each child's performance, ability, learning style, needs, and pace, and then provide a learning experience that is tailored to these unique developmental needs of the child (Prentzas, 2013). AI-powered educational applications, games, and digital platforms, such as paid educational programs (e.g., IXL, https://www.ixl.com; Raz-Kids, https://www.raz-kids.com) and free reading ones (e.g., Epic, https://www.getepic.com; ReadTheory, https://readtheory.org), can personalize or individualize learning experiences for children in various ways, such as by adapting the difficulty level of activities based on the child's individual strengths and weaknesses in performance, providing immediate feedback, and tracking the child's learning progress over time. Chatbots/virtual assistants (e.g. the travel chatbot Octa (https://www.octa.ai) or ChatGPT) can personalize support for learners in a conversational manner by answering questions and providing information and advice. Furthermore, the personalized learning provided through AI-supported services may be viewed as an aspect of child-centered AI-based education (Devi et al., 2022).

Personalized interactive support

Interactive AI-powered smart toys that are designed to serve various functions, such as recognizing the child's speech/questions and responding accordingly, can promote learning that is not found in traditional education. For instance, children (aged four to six) can learn the functions of AI by building, programming, training, and interacting with PopBots (Williams et al., 2019). Children (aged four to five) can foster three types of inquiry literacies (creative, emotional, and collaborative) through playing with AI-interfaced robotic toys (Kewalramani et al., 2021). In addition to providing cognitive support, a social robot can serve as a “friend,” offering emotional support to children as demonstrated in a study by Cañamero and Lewis (2016) on 7-12 year olds with diabetes.

Increased accessibility to broadened learning experiences

AI-powered assistive technologies—such as the speech-to-text, text-to-speech, and automatic speech recognition—can serve as educational support and be made accessible for children who will likely benefit from these services, such as those with special educational needs and those with learning difficulties. For instance, Speechify (https://speechify.com), an AI-powered assistive text-to-speech tool for listening to rather than reading the text, can help enhance student learning and is especially beneficial to those with learning disabilities, such as dyslexia. Intelligent tutoring systems (ITS), that are accessible anytime and anywhere, can offer effective and efficient support in areas, such as language learning (Slavuj et al., 2015; Swartz and Yazdani, 2012). An example of ITS is Duolingo (https://www.duolingo.com), a tool that helps language learners practice reading, writing, and communication skills.

The negative effects of AI in early childhood education

While AI has revolutionized and advanced student learning by providing unique kinds of support services that are not typically found in traditional education, it is critical that we consider its potential pitfalls. We discuss two here.

Overuse of AI

Just because we can does not mean that we should use AI technology excessively. While AI has made access to information more convenient and easier, excessive use can make children develop overdependency, which can conceivably lead to a reduction in or a lack of human interaction. Furthermore, while AI-powered tools may act like a teacher by effectively and efficiently supporting certain aspects of children's learning, they cannot replace the role of human teachers, as well as parents and peers, who provide the kinds of support (such as emotional support and social interaction) that cannot be fully simulated or replicated by AI. Research has consistently attested that human interaction is critical for children to develop the social and emotional skills needed to interact competently with others (e.g. Chen and Adams, 2022; Chen and Badolato, 2023; Chen and O’Donnell, 2022). Thus, children should seekout AI only as a support system when needed, rather than relying on it overly at the expense of turning to their parents, teachers, and peers for academic and emotional assistance.

Misuse of AI

AI-powered technologies can support children's learning and development. However, if misused or used inappropriately, these tools can do more harm than good. For example, if children rely heavily on ChatGPT to generate answers to their homework questions or complete written assignments, it can undermine the educational process and hamper their ability to develop as well as apply critical thinking and problem-solving skills. Relatedly, when children rely on AI to “do the work” for them instead of putting in the effort to learn and understand the work, it can impede their learning and academic success in the long run. Worse, such intention and action can infringe on the integrity and ethical aspects of AI use. For one, the final product may not reflect the children's true learning and understanding. For another, the information provided by AI may be inaccurate, biased, irrelevant, or outdated. At worst, the misuse of AI-generative technologies can lead to negative consequences, such as plagiarism and other forms of academic dishonesty. For example, children may inappropriately copy the ideas generated by AI to pass off as their own.

Applying the POWER principles to guide the appropriate use of AI

While there is no silver bullet for solving the problems of overuse and misuse of AI-powered technologies that are becoming increasingly omnipresent, there are principles that teachers and parents can learn to understand and apply. In turn, they can help guide children in using AI appropriately to potentially maximize its positive effects and minimize its negative effects on these children's learning. We propose five action-driven and conceptually intertwined guiding principles in the acronym POWER (purposeful, optimal, wise, ethical, responsible) and provide examples to demonstrate the how-to aspect of engaging appropriately with AI (see Table 1).

Table 1.

The POWER principles: Concepts, questions to consider, and examples of application.

POWER concepts	Sample questions to consider	Examples of application
Purposeful (to act in a manner that demonstrates clear intention, motivation, and understanding of what one wants to achieve, why, and how to achieve it)	Why do I use this specific AI-powered tool? What positive outcomes/results can this tool help achieve? How can I use this tool with the intention of achieving positive outcomes/results?	Under adult supervision when needed, young children may access the aforementioned child-friendly travel chatbot, Octa, which provides travel information and advice about fun places to visit in any city in the world through its automated messaging-based services.
Optimal (to act in a manner that leverages a technology to its full potential in the most efficient and effective ways to achieve the best outcomes/results)	What can this specific AI-powered tool do best? What functions does it serve best? How can I wield the power of this tool to help achieve the most favorable outcomes/results?	Young children can use PopBots (a hands-on AI tool kit and curriculum) to learn basic AI concepts (Williams et al., 2019). Similarly, young children can leverage Zhorai, a conversational agent and curriculum, to understand how machines learn (Lin et al., 2020).
Wise (to act in a manner that demonstrates thoughtful reflection, good reasoning, and sound judgment by considering various factors, such as advantages, disadvantages, and ethical concerns)	What developmentally appropriate AI-powered tools should I leverage, and why? What potential benefits and limitations/drawbacks should I consider?	The use of intelligent tutoring systems, based on deep learning methods, can wisely offer children individualized constructive feedback and personalize study plans to track their learning performance, progress, and needs. The personalized nature of such AI-powered tools has been found to enhance student learning and educational equity (Ma et al., 2014; Minn, 2022).
Ethical (to act in a manner that is consistent with a set of moral principles or values, such as honesty, integrity, justice, and equity)	How can I use AI-powered technologies to assist in achieving desirable outcomes/results in an ethical manner? What ethical issues should I consider?	Parents, teachers, and children should receive AI ethics training to better understand how to use AI-powered tools ethically, such as critically examining and using the ideas produced by AI merely as a source of guidance or inspiration for developing one's own.
Responsible (to act in a manner that shows good judgment, such as making decisions and choices with safety and accountability in mind, and focusing on doing good and not harm)	How can I use AI-powered technologies in developmentally, culturally, and linguistically appropriate manners? How can I use AI-powered technologies to yield good and not harm?	Under adult supervision when needed, children may use AI-generative tools (e.g., ChatGPT; DALL·E 2, https://openai.com/product/dall-e-2) for learning purposes. However, both adults and children should also learn to be mindful and critical of possible inaccurate, biased, irrelevant, or outdated information, responses, and products provided by these tools.

These POWER principles can serve as a critical aspect of AI literacy, which is defined as the ability to understand, use, assess, and reflect on AI applications (Long and Magerko, 2020; Ng et al., 2021). Young children may already be engaging with AI-powered technologies. However, they may not have acquired AI literacy, including understanding the concepts and principles underlying these technologies and using them in an ethical and responsible manner (e.g. Kong et al., 2021, 2022; Long and Magerko, 2020; Ng et al., 2021; Su and Yang, 2022). To keep pace with the advances in AI use within education, it is particularly imperative that teachers, parents, and children all learn to develop and apply AI literacy appropriately.

Conclusion

While AI capabilities continue to emerge and expand, they will increasingly pervade many realms of society and human life. In education, AI is a powerful catalyst that can transform, personalize, and advance children's learning and development. Yet, AI is also a double-edged sword -- having both upsides and downsides. AI-powered technologies are empowering, but only by understanding their power and limitations can we integrate them more effectively and efficiently in the educational process to enhance teaching and student learning. While AI literacy has made inroads in the education space, the proposed core POWER principles can further provide a critical roadmap for guiding the appropriate use of AI-powered tools to help build a productive partnership between AI and education, thereby benefiting children's learning. In this connection, it would be beneficial to integrate the POWER principles into AI literacy or broadly digital literacy in early childhood education and beyond as they have the potential for relevant and wide-ranging applications across all educational levels and sectors. However, future research is merited to validate these principles and assess their practical implementation by teachers, parents, and children.

Footnotes

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Jennifer J Chen

Author biographies

Jennifer J Chen, EdD, is tenured full professor of early childhood and family studies at Kean University in New Jersey, USA. She has published more than 70 authored/co-authored scholarly works, including three books and many research articles, in the fields of early childhood education and psychology. Her most recent line of research focuses on artificial intelligence in early childhood education.

Jasmine C Lin is a junior in West Essex High School in New Jersey, USA. During her research assistantship, she co-authors articles with Dr. Jennifer Chen in the areas of artificial intelligence, child development, and teaching practices.

References

Abbass

(2021) Editorial: What is artificial intelligence? IEEE Transactions on Artificial Intelligence 2(2): 94–95.

Adams

Pente

Lemermeyer

, et al. (2022) Ethical principles for artificial intelligence in K–12 education. Computers and Education: Artificial Intelligence 4: Article 100131.

Calo

(2017) Artificial intelligence policy: A primer and roadmap. UC Davis Law Review 51(2): 399–435.

Cañamero

Lewis

(2016) Making new “new AI” friends: Designing a social robot for diabetic children from an embodied AI perspective. International Journal of Social Robotics 8(4): 523–537.

Chen

Adams

(2022) Drawing from and expanding their toolboxes: Preschool teachers’ traditional strategies, unconventional opportunities, and novel challenges in scaffolding young children’s social and emotional learning during remote instruction amidst COVID-19. Early Childhood Education Journal. DOI: https://doi.org/10.1007/s10643-022-01359-6.

Chen

Badolato

(2023) Scaffolding social and emotional learning in preschool children from low-income backgrounds: A study of teacher strategies during COVID-19. Education. 3–13. DOI: https://doi.org/10.1080/03004279.2023.2168131.

Chen

O’Donnell

(2022) Facilitating children’s social and emotional development in virtual learning: Implications for in-person teaching. Kappa Delta Pi Record 58(S1): 82–86.

Copeland

(ed.) (2004) The Essential Turing: Seminal Writings in Computing, Logic, Philosophy, Artificial Intelligence, and Artificial Life plus The Secrets of Enigma. Oxford: Clarendon Press.

Devi

Sreedhar

Arulprakash

, et al. (2022) A path towards child-centric artificial intelligence based education. International Journal of Early Childhood Special Education 14(3): 9915–9922.

10.

Jobin

Ienca

Vayena

(2019) The global landscape of AI ethics guidelines. Nature Machine Intelligence 1: 389–399. DOI: https://doi.org/10.1038/s42256-019-0088-2.

11.

Kewalramani

Kidman

Palaiologou

(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.

12.

Kong

Cheung

WMY

Zhang

(2021) Evaluation of an artificial intelligence literacy course for university students with diverse study backgrounds. Computers and Education: Artificial Intelligence 2: Article 100026.

13.

Kong

Cheung

WMY

Zhang

(2022) Evaluating artificial intelligence literacy courses for fostering conceptual learning, literacy and empowerment in university students: Refocusing to conceptual building. Computers in Human Behavior Reports 7: Article 100223.

14.

Lin

Van Brummelen

Lukin

, et al. (2020) Zhorai: Designing a conversational agent for children to explore machine learning concepts. Proceedings of the AAAI Conference on Artificial Intelligence 34(9): 13381–13388.

15.

Long

Magerko

(2020) What is AI literacy? Competencies and design considerations. In: CHI ’20: Proceedings of the 2020 CHI conference on human factors in computing systems, Honolulu, HI, USA, 25–30 April 2020, pp. 1–16. New York: Association for Computing Machinery.

16.

Adesope

Nesbit

, et al. (2014) Intelligent tutoring systems and learning outcomes: A meta-analysis. Journal of Educational Psychology 106(4): 901–918.

17.

McCarthy

(2007) From here to human-level AI. Artificial Intelligence 171(18): 1174–1182.

18.

McCarthy

Minsky

Rochester

, et al. (1955) A proposal for the Dartmouth summer research project on artificial intelligence. Available at: http://raysolomonoff.com/dartmouth/boxa/dart564props.pdf

19.

Minn

(2022) AI-assisted knowledge assessment techniques for adaptive learning environments. Computers and Education: Artificial Intelligence 3: Article 100050.

20.

National Association for the Education of Young Children

(NAEYC)

(2022) Developmentally Appropriate Practice in Early Childhood Programs Serving Children from Birth through Age 8. 4th ed. Washington, DC: National Association for the Education of Young Children.

21.

DTK

Leung

JKL

Chu

SKW

, et al. (2021) Conceptualizing AI literacy: An exploratory review. Computers and Education: Artificial Intelligence 2: Article 100041.

22.

Prentzas

(2013) Artificial intelligence methods in early childhood education. In: Yang

X-S

(ed.) Artificial Intelligence, Evolutionary Computation and Metaheuristics: In the Footsteps of Alan Turing. Berlin/Heidelberg, Germany: Springer, pp. 169–199.

23.

Slavuj

Kovačić

Jugo

(2015) Intelligent tutoring systems for language learning. In: MIPRO 2015: 38th international convention on information and communication technology, electronics and microelectronics, Opatija, Croatia, 25–29 May 2015, pp. 814–819. Rijeka: Croatian Society for Information and Communication Technology, Electronics and Microelectronics. Available at: https://doi.org/10.1109/MIPRO.2015.7160383

24.

Yang

(2022) Artificial intelligence in early childhood education: A scoping review. Computers and Education: Artificial Intelligence 3: Article 100049.

25.

Swartz

Yazdani

M (eds)

(2012) Intelligent Tutoring Systems for Foreign Language Learning: The Bridge to International Communication. Berlin/Heidelberg, Germany: Springer.

26.

Vartiainen

Tedre

Valtonen

(2020) Learning machine learning with very young children: Who is teaching whom? International Journal of Child–Computer Interaction 25: Article 100182.

27.

Vygotsky

(1978) Thought and Language. Cambridge, MA: MIT Press.

28.

Wertsch

(1984) The zone of proximal development: Some conceptual issues. New Directions for Child and Adolescent Development 23: 7–18.

29.

Williams

Park

Breazeal

(2019) A is for artificial intelligence: The impact of artificial intelligence activities on young children’s perceptions of robots. In: CHI ’19: Proceedings of the 2019 CHI conference on human factors in computing systems, Glasgow, UK, 4–9 May 2019, pp. 1–11. New York: Association for Computing Machinery.