Author Institution │ Beijing Normal University, Faculty of Education, College of Educational Technology
Source │Journal of Information Technology Education for Primary and Secondary Schools 2024, Issue 10
Artificial Intelligence and Intelligent Agents
In the field of artificial intelligence, intelligent agents, also known as “Autonomous Agents,” are adaptive systems capable of completing goal-oriented tasks. Since the establishment of the discipline of artificial intelligence in the 20th century, the design and implementation of intelligent agents have remained a core objective for researchers in the field. A key characteristic of intelligent agents is their autonomy: they can control their behavior and internal states, make decisions, and execute corresponding actions based on changes in the external environment and their goal tasks, all without direct human intervention, thereby affecting or even altering the external environment. Moreover, intelligent agents must be able to engage in effective communication with humans to better serve their needs.
1. Intelligent Agents Based on Traditional Artificial Intelligence
Intelligent agents in the field of artificial intelligence first need to possess the ability to perceive the external environment to gather necessary information, thus understanding the basic conditions and dynamic changes of the external environment. Based on the collected external environmental information, intelligent agents can solve specific problems in goal tasks through deterministic or non-deterministic logical reasoning. Additionally, intelligent agents can utilize data collected from the external environment to automatically deduce objective rules within the data through machine learning, thereby improving their effectiveness and efficiency in solving goal tasks. During this process, intelligent agents can establish memory and retrieval mechanisms to support the storage and retrieval of data and past experiences, further enhancing their performance. On this basis, intelligent agents can weigh pros and cons using specific mathematical models and algorithms, make rational decisions regarding goal tasks, and translate these decisions into actual effects and impacts on the external environment through execution mechanisms, ultimately completing the goal tasks.
The design of intelligent agents typically relies on various traditional artificial intelligence technologies, including but not limited to speech recognition, computer vision, knowledge representation and reasoning, machine learning, natural language processing, and robotics. The specific implementation of intelligent agents can take the form of software, such as mobile personal assistants, e-commerce recommendation systems, and online customer service robots; or hardware, such as autonomous vehicles, industrial robots, and service robots.
2. Intelligent Agents Based on Multimodal Large Models
With the rapid development of generative artificial intelligence technology, particularly the emergence of multimodal large models (hereinafter referred to as “large models”), new opportunities arise for the design and construction of intelligent agents. Large models refer to artificial intelligence models capable of processing and interpreting multimodal data inputs, including text, images, audio, and video. Large models represented by GPT-4 typically possess hundreds of billions of parameters and exhibit outstanding performance in natural language processing and audiovisual analysis across multiple tasks. In recent years, researchers have begun to attempt to build intelligent agents that rely on large models as their core foundational support. Compared to traditional intelligent agents, large model-based intelligent agents have significant advantages and characteristics, particularly the following core capabilities.
(1) Multimodal Perception Capability.Large models can perceive various modalities of data, such as images, speech, and text, achieving independent or integrated perception functions. Therefore, intelligent agents built on large models exhibit more intelligent multimodal perception capabilities, allowing them to comprehensively understand their external environment. These agents can not only perceive visual and auditory information but also integrate information from different modalities, achieving a level of perception closer to that of humans.
(2) Reasoning and Planning Capability.Large models possess strong logical reasoning abilities, allowing designers to set “prompt information” to stimulate the model for deeper thinking, forming coherent “chains of thought” or systematic “trees of thought.” Thus, intelligent agents based on large models can decompose complex tasks into a series of actionable sub-tasks. Through multi-step logical reasoning, they can autonomously plan a series of actions to accomplish goal tasks, efficiently exploring and implementing strategies to solve complex problems.
(3) Learning and Decision-Making Capability.Large models, due to their massive parameter scale and complex artificial neural network architecture, accumulate a comprehensive understanding of the objective world through training on vast multimodal data, providing a solid knowledge base for effective decision-making. Additionally, large models utilize techniques such as “fine-tuning” and “retrieval-augmented generation” to deeply learn specific features and knowledge for particular domains or tasks, reducing the occurrence of “hallucinations” (i.e., erroneous or inaccurate information). Therefore, intelligent agents based on large models can provide more comprehensive and in-depth information, ensuring the quality and reliability of the learning and decision-making processes.
(4) Multi-role Interaction Capability.Large models have the ability to understand and capture contextual information in multi-turn dialogues, analyzing conversational situations and user intentions, thus generating logically coherent and appropriate responses. Consequently, intelligent agents built on large models can interact more effectively with human users, external environments, and other intelligent agents. In human-computer interaction, intelligent agents can collaborate with human users while providing high-quality user experiences. In interactions with other intelligent agents, different agents can engage in effective debates and collaborations based on their respective roles and functions, collectively advancing the efficient completion of complex tasks.
(5) Memory and Evolution Capability.Large models can utilize external memory and retrieval mechanisms in artificial intelligence systems to effectively store and retrieve specialized knowledge. Thus, intelligent agents based on large models can reflect on and summarize external memories and specialized knowledge, achieving a level of autonomous thinking and self-improvement similar to that of humans. This self-evolution capability not only enhances the adaptability and flexibility of intelligent agents but also provides possibilities for continuous optimization, maintaining their advantages and efficiency in ever-changing external environments and while completing complex tasks.
Intelligent Agents for the Education Sector
The deep integration of artificial intelligence and education has given rise to intelligent agents specifically designed for the education sector, known as educational agents. Due to the diversity of educational scenarios and the complexity of service targets, the construction of educational agents must meet the unique needs of the education field and exhibit specialized characteristics and functions distinct from intelligent agents in other vertical domains.
1. Setting Educational Tasks
Educational agents need to be able to set educational scenarios, educational needs, and educational roles based on target educational tasks. The setting of educational scenarios provides the background and environmental information for educational tasks, such as project-based learning scenarios centered on students or traditional classroom teaching settings; the setting of educational needs provides specific goals and descriptive information for educational tasks, such as driving questions for project-based learning or evaluating teachers’ classroom teaching capabilities; and the setting of educational roles assigns specific role information needed for educational tasks, such as peer students or research experts.
2. Planning Educational Tasks
Under the guidance of established educational tasks, educational agents need to achieve autonomous task planning. First, educational agents can fully utilize the reasoning and planning capabilities of large models to autonomously conceive and design preliminary solutions based on the scenario and needs information set in the educational tasks. On this basis, educational agents will decompose the generated overall plan into multiple actionable sub-tasks. Educational agents must continuously monitor the actual implementation effects of each sub-task and receive feedback from educational users. If the goals or expected results are not met, educational agents need to dynamically adjust the solutions and sub-tasks to ensure the achievement of educational tasks.
3. Implementing Educational Tasks
For each planned sub-task, educational agents can initially utilize their own learning and decision-making capabilities based on large models for effective resolution. For sub-tasks that exceed the direct handling capabilities of the agents, educational agents need to proactively call external third-party tools or query knowledge bases online for resolution. For example, for subjects like mathematics that require precise calculations, educational agents can actively access third-party calculation tools to ensure the accuracy and efficiency of calculations. To provide the latest educational resources and data, educational agents can also access professional educational resource public service platforms online to obtain the required information.
4. Interaction with Educational Users and Self-Evolution
In the process of completing target tasks, educational agents can interact with human users, other intelligent agents, and the educational environment, continuously evolving themselves in the process. In interactions with people, intelligent agents can gain insights into the needs of different educational users and roles, providing a personalized and multimodal human-computer interaction experience; in collaboration with other intelligent agents, they can advance the completion of educational tasks through discussions and debates; and in interactions with the educational environment, educational agents can perceive and respond to environmental changes, optimizing teaching settings. Furthermore, after various interactions, educational agents can reflect and continuously improve through their memory and evolution mechanisms.
5. Building an Educational Knowledge Base
The achievement of educational tasks typically requires high accuracy and interpretability to enhance the trust of educational users. Therefore, educational agents often need to rely on professional educational knowledge and data during the planning and execution of educational tasks. To meet these requirements, a local specialized educational knowledge base needs to be constructed for educational agents, also encompassing personalized information about users. Educational agents can efficiently utilize information resources from the local educational knowledge base through techniques like “retrieval-augmented generation,” integrating this information into generated answers or solutions to enhance the accuracy and reliability of the educational information output. Additionally, through continuous interaction with the local knowledge base, educational agents can learn and adapt, expanding their knowledge boundaries and improving their capability to handle various educational tasks. This ongoing learning and adaptation process helps the agents better understand educational content and provide more personalized and targeted educational services.
Applications of Educational Agents
The research and development of educational agents are still in their infancy, but they have already begun to demonstrate significant application potential and important roles in typical educational scenarios such as classroom teaching, educational evaluation, and teacher research.
1. Classroom Teaching
In classroom teaching scenarios, teachers and intelligent agents can form collaborative partnerships to jointly advance the implementation of teaching activities. Intelligent agents can assume diverse roles and functions in the classroom, providing multidimensional support to teachers. For instance, in project-based learning teaching models, learners rely on continuous support from teachers and peers to achieve project goals. In this process, educational agents can serve as “assistants” and “peers,” collaborating with teachers and learners to fully participate in all stages of project-based learning. This includes proposing personalized questions, collaboratively designing project plans, and co-creating project outcomes. Furthermore, educational agents can automatically recognize individual students’ emotional states during the collaboration process and provide corresponding emotional support and interactive feedback. Human-computer collaborative teaching supported by educational agents can not only improve teaching efficiency but also promote students’ active exploration and teamwork, playing a supportive role in future classroom teaching.
2. Educational Evaluation
In educational evaluation, educational agents can leverage their multimodal perception capabilities and memory functions to promote the transition from experience-dependent subjectivity to data-driven objectivity, thereby enhancing the objectivity and fairness of educational evaluations. For example, during the evaluation and presentation phase of project-based learning, educational agents can utilize detailed process information stored in their memory modules about group collaboration to conduct thorough process evaluations of project implementation. At the same time, educational agents can implement teacher evaluations and peer assessments from different perspectives of “teachers” and “peers.” Educational agents can also pre-generate corresponding process and outcome evaluation rubrics based on personalized driving questions and project plans.
3. Teacher Research
With the assistance of educational agents, teachers can collaboratively prepare lessons and conduct research activities. Educational agents can first provide teachers with teaching resources, including teaching plans and multimodal teaching materials, and simulate and display classroom situations, assisting teachers in iteratively optimizing preparation materials using external tools and resources to create teaching content that better meets students’ actual needs. After teachers complete teaching activities using these resources, educational agents can quantitatively analyze the objective dimensions of teachers’ teaching capabilities based on their perceptions and memories of classroom implementation, providing detailed analysis reports. In research activities, educational agents can also assume the roles of experts from various subject backgrounds, performing lesson observation and evaluation tasks to conduct multidimensional analyses and evaluations of the teaching process. Teachers can interact and even debate with educational agents to improve their teaching methods and further optimize the teaching process.
Conclusion and Outlook
The construction of intelligent agents has long been a pursuit in the field of artificial intelligence. The development of educational agents requires a deep understanding of educational resources, learner characteristics, and teaching processes, supported by solid educational theories and learning sciences. Educational agents should possess the ability for continuous learning and self-improvement, achieving autonomous evolution and effective interaction among multiple intelligent agents through engagement with educational stakeholders. At the same time, the design and application of educational agents must comprehensively consider their far-reaching impact on education and potential challenges. Designers should focus on enhancing the reliability and credibility of educational agents, ensuring they provide unbiased and fair educational services, and thoughtfully consider their influence on the values and ethics of learners.
END
AICFE
Join the Discussion Group Contact Us
■ Basic Education Leapfrog Development Experiment Discussion Group 1
Please add Teacher Ji’s WeChat (Jll_0405), note your name and profession, and request to join the group.
■ “High-Precision and Cutting-Edge” Future Education Discussion Group 3
[Please add Teacher Guo’s WeChat (guojl0330), note your name and profession, and request to join the group.]