Why I Am NOT Creating An Educational AI Chatbot
Introduction
In the current landscape of educational technology, the allure of AI-powered chatbots and storybots is undeniable.
However, I am not building one because early readers deserve a better solution.
These seemingly magical chatbots promise a seamless interweaving of artificial intelligence into the fabric of learning, suggesting a future where education is both accessible and technologically advanced.
Unfortunately, in the realm of EdTech, particularly in the foundational stages of reading and literacy, the prevalent model of AI educational chatbots falls far short of addressing the nuanced complexities of teaching young learners. The chatbots are often merely sophisticated wrappers around Large Language Models (LLMs) such as ChatGPT or Gemini. While they dazzle with their ability to mimic human conversation, their effectiveness pales in comparison with the intricate, personal journey of learning to read.
My belief in AI as a cornerstone of educational innovation remains unwavering; however, I posit that our focus has narrowed too much on the standalone capabilities of what we are currently describing as AI.
This perspective risks overlooking the broader spectrum of needs in the educational landscape. The practical effectiveness lies not in what AI can do in isolation but in how it can be synergistically integrated with other essential components of the learning experience. It's about crafting a holistic educational environment where AI is a pivotal element, enhancing and being enhanced by other pedagogical tools and methods.
This comprehensive approach promises not only to meet but to exceed our educational aspirations, harnessing the full scope of possibilities that AI can offer when thoughtfully integrated: education at scale that is hyper-personalized for each reader.
When I consider the optimal learning environment for children embarking on their reading journey—the limitations of chatbots become starkly apparent. Reading, particularly in its nascent stages, is an intensely personal, multifaceted process that demands more than a mere mimicry of human conversation; it must have the situational awareness of a fantastic teacher and the personal understanding of a caring parent.
Consider this scenario: an early reader named Alex is reading a fun story when he encounters a new word, “Butterfly,” and he has some difficulty with it. How does the chatbot know there is an issue? And if it does perceive a difficulty, how does it determine the underlying issue? An LLM Chatbot may prompt him to retry, offering mere repetition or moving on without grasping the true nature of his struggle.
In contrast, my new product, Leonardo, with its gaze and audio analysis, discerns Alex's specific hesitation because of its situational awareness. Is Alex “stuck” on this word, or is he regressing across the sentence? Is Alex looking at the screen, or has he been distracted or interrupted? Leonardo can then respond appropriately by taking any number of actions: slowing the pace, highlighting "butterfly," and simplifying the surrounding text to ease the cognitive load. Leonardo can also pull Alex aside, like a teacher or a caregiver would, and work on the word through an engaging, multisensory game that teaches Alex how to piece the word together, continuing to adjust to Alex's individual needs.
Additionally, Leonardo reinforces similar words in subsequent text (in this session or future ones) to give Alex further practice with the specific lesson. Ultimately, Leonardo works through solving the underlying issues to foster confidence and mastery in a supportive, adaptive manner rather than only offering impersonal, repetitive, crowd-sourced interactions.
An effective platform for early readers must surpass the mimicry of human conversation; it must possess the awareness of a gifted teacher and caregiver. This is why we have collaborated with educational and children's media experts to develop a set of 5 guiding principles:
1. One Cannot Learn to Read Text Via Text Alone
As adults, it is easy for us to forget the challenges of reading. To understand the alphabet is to decode a series of abstract symbols, assigning them to sound and meaning, which is a task of immense proportions for a young mind. This image exemplifies the experience of a first introduction to the alphabet:
This initial hurdle in literacy is about more than just recognizing letters but also understanding their roles in word formation. It's a journey from seeing arbitrary shapes to extracting meaning, context, and narrative from them. Expecting a child, especially one at the cusp of literacy, to navigate this path through textual-based (even text-to-speech) interactions with a chatbot overlooks the essence of early reading. Text is not the sole medium to teach text; the personal and pedagogical approach must be as multidimensional as the skill it aims to impart.
2. Sensory Awareness is a Must-Have
Chatbots have no situational awareness. They don’t know anything except what someone on the other end has typed, and they can only respond to that. They have no idea if the child is engaged with the experience or where on the screen they are fixated, their disposition, their performance, etc.
At Leonardo Learning, we transcend chatbots' limitations through a multimodal approach that harnesses voice, touch, and vision. Understanding a current situation (performance, engagement, etc.) is the key to understanding the context of each session and generating content specific to it.
This methodology also acknowledges that children learn through diverse interactions beyond the written word. Leonardo creates more effective learning experiences by leveraging multiple senses, catering to the varied ways children perceive and interact with their environment. This holistic engagement fosters a deeper, more intuitive understanding of reading, making each learning moment resonate more intensely.
3. The Importance of Personalized Engagement
If a reader is not engaged, nothing else matters.
The heart of Leonardo Learning lies in its ability to generate content dynamically, tailored not only to the child's current reading level but also woven with threads of their personal interests. This individualized approach ensures that learning materials are not merely educational but fundamentally engaging, encouraging children to connect with the content on an individual level and even see themselves within it.
Unlike chatbots operating within constrained parameters, Leonardo adapts in real-time, crafting stories as unique as the child reading them. This personal touch transforms learning from a mundane task into an amazing adventure, where each page turn reveals content that speaks directly to the child's curiosity and interests.
4. Real-Time Adaptivity and Feedback are Key
The essence of teaching, particularly in the foundational stages of learning to read, lies in the ability to adapt—to recognize the moment when a concept clicks or to identify when a child is struggling. Leonardo's technology mirrors this adaptive quality of an attentive educator. Through real-time analysis of the child's interactions, Leonardo adjusts its teaching strategies just in time to meet the child’s needs, generating real-time content in the background that is both relevant and timely.
This immediate adaptivity ensures that learning is efficient and profoundly aligned with the child's personal learning journey, a feat no chatbot, with its turn-based interaction model, can achieve.
5. We must be careful about HOW we teach
This is the most crucial point.
As someone who considers himself highly creative and also highly technical, I have been haunted my entire adult life by the work of Viktor Lowenfeld and Lambert Brittain, who illustrated brilliantly the unforeseen consequences of representing things efficiently but not effectively and the subsequent impact it has on a young mind. When they asked children to draw a bird, they received wonderful and imaginative depictions like this one (fig. 1):
Fig. 2 (Original)
However, upon exposing the children to a simple math worksheet (Fig. 3), asking them to color one bird blue, the children were unintentionally imprinted with a different, less creative representation.
Fig. 3 (Math Worksheet Image)
When subsequently asked to draw birds, they created these tragic representations (Fig. 4 post worksheet):
Fig. 4 (Post Worksheet)
Illustrations are from Viktor Lowenfeld and W. Lambert Brittain. Creative and Mental Growth, 6th ed. 1970. Macmillan, NY. p. 109
It is supremely important that we be mindful of the potential impact of our teaching methods and avoid the inadvertent instillation of restrictive stereotypes or simplified viewpoints.
Until AI Chatbots can make this nuanced human-centric distinction, I cannot entrust them with the proper educational choices for all of our children.
This is why Leonardo uses well-thought-out baseline stories crafted by some of the world's most respected educational experts and only uses AI to personalize these texts for each reader's passions, preferences, and performance. This is a much safer and more predictable approach.
Leonardo Learning has taken a more humane approach, transcending the simplistic model of chatbots to offer a rich, personalized, and profoundly aligned learning experience for young learners.
As we look to the future of education, it is clear that the path lies not in the mere parroting of human conversation but in the thoughtful integration of technology within the timeless principles of effective teaching.
Leonardo Learning is not just another Edtech chatbot. It is a reimagining of what it means to learn, an invitation to embark on a journey of discovery as limitless as the potential within every child.
