Two days later, this question was systematically addressed at both the technical and classroom levels. On November 11, at the 9th China Education Technology Innovation Development and Application Achievements Conference held in Wuxi, Jiangsu, Dr. Deng Shujun, Chief Scientist of Galaxy Smart Learning, gave a keynote speech titled "Let Every Child Learn, Learn Well, and Love Learning - Exploration and Practice of AI Supporting Scalable Personalized Teaching," detailing the technical architecture and school scenario implementation logic of the company's product "Xiaolü Ai Xue." 
In Deng Shujun's view, large models can quickly generate problem-solving steps, but they may not be able to answer "why the student doesn't understand." As general models become infrastructure, the true dividing line for educational AI lies in its ability to deeply understand the learning process - grasping the logic of subjects, identifying individual cognitive differences, and transforming technology into verifiable learning outcomes in real classrooms.
Galaxy Smart Learning was founded by Chen Lin in December 2024. The core members of Galaxy Smart Learning have previously worked in ByteDance's education business and the related teams of the Doudou model. The team has continuously focused on "student self-learning" and intelligent education systems, accumulating experience in the development of self-learning products, curriculum research and teaching, as well as training in long-term memory, problem-solving, and teaching with large models. Among them, Chief Scientist Deng Shujun previously served as the head of Qinghua and Peking Online School and the algorithm team at Dali Education, and participated in the development of Doudou Ai Xue and He Ma Ai Xue. 
Dual-core technology foundation: enabling AI to both understand students and inspire thinking
Although general large models have strong general capabilities, they may still produce hallucinations when dealing with rigorous subject logic. To meet the needs of educational scenarios, Galaxy Smart Learning adopts a "dual-core" technical architecture combining the Doudou model with its self-developed intelligent learning engine, deeply integrating general intelligence capabilities with educational professional algorithms.
The general large model mainly handles tasks such as language understanding, content generation, and basic reasoning. The team has conducted specialized optimization for educational scenarios, including the construction of problem-solving thought chains, alignment with Socratic-style prompting, and the development of Agent capabilities based on long-term memory. By breaking down key steps in the problem-solving process, the model can pause at appropriate stages and ask prompting questions, guiding students to gradually understand knowledge and discover cognitive blind spots. 
According to the introduction, the self-developed intelligent learning engine is based on learning science and cognitive measurement analysis, integrating technologies such as deep knowledge tracing (DKT), item response theory (IRT), cognitive diagnostic models (CDM), and knowledge space theory (KST). By analyzing students' responses, questions, and correction records, it can determine what knowledge the student has already mastered, what weaknesses they currently have, and what content is suitable for the next step, providing a basis for learning situation diagnosis, ability assessment, matching learning difficulty, and subsequent task planning.
The two form a complementary system, with the general large model responsible for knowledge explanation, prompting questions, and process guidance, while the self-developed intelligent learning engine is responsible for learning situation diagnosis and learning path judgment. As Dr. Deng Shujun summarized in his speech, the former should "guide well," and the latter should "calculate accurately." Building on this, Galaxy Smart Learning further extends the Agent capabilities to student learning planning and teacher teaching support.
The planning intelligent agent on the student side relies on the long-term memory of the large model, combining with the student's long-term learning situation, to recommend the next learning task and explain the basis for the recommendation. The teaching assistant intelligent agent on the teacher side extracts specific effort and progress clues from the student's learning records, providing a reference for teachers to conduct personalized guidance.
The AI learning loop further connects students' learning behaviors, teachers' instructional guidance, and school feedback in a continuous manner. Students' learning performance enters the next round of learning situation analysis and task arrangement, and teachers can adjust their teaching strategies based on the feedback, allowing AI capabilities to continuously influence real learning processes.
The core of this technical architecture is to combine subject knowledge, cognitive measurement, learning situation analysis, and the interaction guidance of large models, and to continuously optimize future support through ongoing learning feedback.
From answering questions to guiding thinking, AI needs to enter a complete learning process
For educational AI, technical capabilities ultimately need to be transformed into specific product interactions, implemented in understanding the student's learning process, and continuously validated in terms of learning outcomes.
Currently, many AI education products still focus on question-answering. Students input questions, AI generates explanations, and the learning process often ends with obtaining an answer. But true learning also involves understanding problems, organizing thoughts, practicing applications, and checking mastery, all of which collectively determine whether knowledge can truly be transformed into students' own abilities.
In the design of Xiaolü AiXue, the team emphasizes "letting technology go behind the scenes and letting students take the lead," and has designed specific functions around the complete learning process: AI provides support at the right time, allowing students to personally experience selection, trial, error, adjustment, and verification, gradually learning to recognize and manage their own learning during the process of mastering knowledge. The system supports students in asking multiple rounds of questions and follow-up questions around the learning content, connecting knowledge understanding, practice application, and effect evaluation. 
Dr. Deng Shujun gave an example in his sharing: when the height of a cylinder remains unchanged and the radius of the base is doubled, the volume should be four times the original, but some students might mistakenly think the volume only doubles. To address such understanding biases, Xiaolü AiXue introduces dynamic concept experiments and Socratic prompting: students can intuitively observe the impact of radius changes on the volume of a cylinder through a slider, while AI asks layered questions around key steps such as the volume formula and changes in base area, guiding students to derive conclusions on their own.
"These prompts build a 'ladder' for students to understand knowledge, helping them find the first step they can take and think independently at critical moments, answering step by step." As understanding deepens, the system will appropriately remove support, testing whether students can answer correctly, explain their reasoning, and transfer their learning to new problems through independent answering, explaining their thought process, and variant practice. "The team hopes that students will gain a sense of 'aha!' and achievement by going through the process of overcoming difficult problems."
In cultivating students' interest and confidence, the team combines deep knowledge tracking with the theory of the zone of proximal development, using models to predict the probability of students answering the next question correctly, and matching learning tasks with moderate challenges. Dr. Deng Shujun introduced in his speech that the system selects questions with a predicted correct answer rate of about 70% to 80%, allowing students to achieve progress through effort.
"A small breakthrough achieved by oneself can become a 'spark' for a child to rebuild confidence." Dr. Deng Shujun said that the system records students' learning performance at different stages, presenting specific progress compared to the previous day or week, making minor efforts and breakthroughs visible. Learning teams encourage peers to motivate each other and tackle challenges together, while teaching assistants provide timely and targeted encouragement for teachers. "The team hopes that children will feel 'this step was taken by myself,' gradually accumulating confidence that 'I can learn,' thus being more willing to face the next unknown and take another step forward." 
For school scenarios, a single AI tool is difficult to form a complete teaching solution. A student may quickly complete exercises with the help of AI, but if they do not truly understand the knowledge, efficiency improvement does not necessarily mean an improvement in learning quality.
Xiaolü AiXue is positioned as a human-AI co-education classroom solution for schools, providing personalized learning support to students by recording and analyzing their learning process, and assisting teachers in understanding students' knowledge mastery and individual differences. Teachers continue to take on the roles of teaching design, key knowledge explanation, learning guidance, and emotional support, and adjust teaching strategies based on AI feedback.
Teachers can understand students' learning progress, common questions, and correction situations through the learning situation dashboard, conducting focused lectures, tiered guidance, and individual tutoring. At the school management level, the principal's end aggregates teaching process data, providing a basis for teaching observation, reflection, and management. Schools incorporate self-directed learning, independent testing, and targeted remedial learning into daily courses and schedules, with test results then fed back into the next round of learning situation analysis and task assignment, forming a teaching loop of "teacher-led, student-initiated, and AI-supported."
The real challenge of AI education is how to scale personalized learning.
Personalized learning is not a new concept. For a long time, teaching according to individual differences has been a goal of education, but in traditional classrooms, teachers' time and energy are limited, making it difficult to continuously track each student's learning process, let alone provide timely and effective personalized guidance for everyone.
The possibilities brought by AI lie in using learning process data and intelligent analysis to reduce the cost of identifying student differences and providing personalized feedback, making parts of the work that were originally highly dependent on teachers' personal experience more timely and sustainable. However, moving from individualized tutoring for a single student to school-scale application still faces many challenges.
According to the introduction, in 2022, core members of Galaxy Smart Learning participated in the on-campus exploration of Heima AiXue in Anhui Fuyang Qinqing School during the period of Dali Education. In the long-term contact with frontline teachers and students, the team found that students' learning difficulties not only manifest in knowledge mastery but also include aspects such as learning methods and learning confidence. This frontline practical experience also influenced the team's later design of AI educational products, making the team more focused on how students understand knowledge, master methods, and whether they can build confidence in continuous progress.
During the large-scale implementation process, differences exist among schools in terms of teaching conditions, course progress, and student foundations. Whether teachers can integrate AI into daily teaching, whether the product can run stably, and whether learning feedback can be truly converted into teaching actions all affect the actual application effect.
Taking Guizhou Nanyong Hongxing High School as an example, the team gradually integrated functions such as recitation dictation, test paper creation and distribution, online testing, AI grading, and learning situation feedback into the school's existing class types, including morning reading, regular classes, time-limited practice, and evening study sessions. This helped teachers to promptly understand students' learning status and conduct explanations for common issues. This continuous co-creation based on real classrooms also allowed the AI system to move beyond being a single learning tool and further integrate into the school's curriculum organization, teaching feedback, and daily management.
From technical solutions to real classrooms, educational AI is entering a deeper stage of application. Currently, Xiaolü Aixue, under Galaxy Intelligent Learning, has been implemented in schools of different regions and types. According to Galaxy Intelligent Learning, the product has served and cooperated with more than 500 schools, covering 24 provincial administrative regions. In recent months, Galaxy Intelligent Learning has gradually disclosed cases such as Anhui Fuyang Qinqing School, Capital Normal University Affiliated Huilongguan Yuxin School, Guizhou Nanyong Hongxing High School, Qingdao Guoke High School, Henan Xincai Lantian High School, and Inner Mongolia Siziwangqi Kuangao Experimental School, showcasing the exploration of human-AI co-education models in different school contexts.
From knowledge acquisition to ability cultivation, the value of learning needs to be re-evaluated in the AI era.
The AI era requires a re-examination of the value of learning.Large models are changing the way knowledge is acquired. If the first stage of educational technology focused more on knowledge acquisition and learning efficiency, then the next change brought by AI might be a re-examination of learning itself.
In the past, students spent a lot of time searching for materials, understanding concepts, and obtaining answers; now, AI can quickly provide information, explain knowledge, and even help complete complex tasks. As the threshold for knowledge acquisition continues to lower, the value of education also needs to shift from knowledge transmission to ability cultivation.
Therefore, the evaluation criteria for AI educational products also need to change: it cannot only focus on answer speed, number of exercises, and frequency of interaction, but also pay attention to whether students can truly understand knowledge, form effective learning methods, and gradually develop the ability to independently solve and actively explore problems, and build interest and confidence in learning through continuous progress.
This is the reason why Galaxy Intelligent Learning has set "learning, learning how to learn, and loving to learn" as its educational goals, and this goal is also related to the founder Chen Lin's decision to re-enter the education industry for entrepreneurship. Chen Lin reviewed his thoughts on education in the AI era in his article. In his view, facing the technological changes brought by AI, education needs to help children develop the ability to deal with uncertainty and solve real problems. At the end of the article, he wrote: "This era has already seen many people making AI stronger and stronger, while I want to help people become better and better..." 
When AI enters the classroom, it ultimately points to human growth: enabling students, under the guidance of teachers and the support of technology, to gradually develop the ability to understand the world, make independent judgments, and solve problems, growing into individuals capable of independently facing the unknown world. And when technology no longer urgently proves how smart it is, but truly understands each child's cognitive blind spots, ignites their intrinsic drive for self-exploration, and forms long-term collaboration with the school's educational practices, educational AI can truly realize its value.
