Enterprise AI Analysis

2 Artificial Intelligence and Computational Thinking

2.1 The connotation and elements of computational thinking

Computational Thinking was first introduced to China in 2006 by Professor J.M. Wing of Carnegie Mellon University in the United States. He defined its meaning as the use of mathematical principles (ideas and methods) to explore, construct, analyze, predict, control, and more. This theory has multiple advantages, the most important of which is to transform abstract concepts into practical applications, convert basic knowledge into practical operations, transform human intelligence into computer intelligence, combine mathematical logic with practice, transform rational thinking into practical actions, and be applicable to various scenarios. In 2008, ACM published the "CC001 Computer Science Teaching Guidance Draft" emphasizing the core concept of computer technology science teaching, which aims to cultivate students' innovative and practical abilities. Shao Bin and Looi C K delved into the relationship between this concept and practice[4][5]. Chen Juanjuan emphasized that the three fundamental scientific and technological thinking models of calculation, empirical research, and theory constitute the basis for exploring the world, and cross-border co-operation has become the most important driving force in this process [6].

However, Professor Zhou Yizhen's statistical ideas and concepts are relatively broad and abstract. There are still many areas worth exploring in depth to grasp the essence of statistical thinking and focus on cultivating students' statistical thinking. Some teachers at home and abroad have conducted a series of studies on how to cultivate students' statistical thinking in the actual teaching process[7][8][9]. As early as 2010, the “Curriculum Strength Structure and Talent Cultivation of Computer Science and Technology Majors in Higher Education Institutions" formulated by the National Education Ministry's Computer Science and Technology Guidance Conference proposed that the disciplinary foundation of electronic computer personnel generally includes statistical thinking ability, algorithm design and analysis ability, programming design and implementation ability, and comprehensive information system ability. This article extracts from the above definition and discusses how to train students' computational thinking in the educational process of artificial intelligence, in order to optimize their professional qualities and creativity.

2.2 The Relationship between Artificial Intelligence and Computational Thinking

One of the goals of artificial intelligence is to create intelligent machines or systems to complement and supplement human intelligence, and computational thinking is the key to the development process of artificial intelligence systems. In terms of the relationship between artificial intelligence and computational thinking, some researchers believe that the process of using artificial intelligence and computational thinking to solve problems has similarities and involves the same elements. For example, Sirapahote found that computational thinking elements such as abstraction, decomposition, and logical reasoning are also necessary for solving problems in the field of artificial intelligence[10]. The thinking processes of both are based on the abstraction process to implement algorithms and apply them to problem solving through modeling. Tamborg pointed out that there are three common elements between artificial intelligence and computational thinking: agency, phenomenon modeling, and abstract concepts that go beyond concrete examples[11]. Asunda believes that specific artificial intelligence related concepts should be incorporated into Brennan&Resnick's computational thinking framework to form a thinking framework that helps understand the process of artificial intelligence development[12]. Some researchers also believe that the current emphasis on computational thinking elements is difficult to apply to solving problems in the field of artificial intelligence, because with the advancement of artificial intelligence technology, the thinking processes of the two are no longer at the same level. As Shamir pointed out, artificial intelligence is "automated automation", which is the process of generating automated solutions through automation[13]. The computational thinking involved is different from traditional computational thinking (decomposition, pattern recognition, abstraction, and algorithm design). Chen Zan'an proposed that in the era of artificial intelligence, computational thinking is different from the skills in the computer field and the way of thinking of applying computation and computational models, but requires dynamic participation and construction of innovative computational models in human-machine collaboration[14]. Pierce believes that modern development platforms, libraries, engines, and cloud services are automating more and more system development activities, containing higher levels of abstraction[15]. The ability to explain intelligent systems using computational thinking defined by reductionism and determinism is weakening.

In summary, it can be found that computational thinking is a fundamental characteristic of artificial intelligence, and the requirements for computational thinking in artificial intelligence are different from the simple requirements in general information technology application fields. From a "procedural" perspective, the computational thinking in artificial intelligence also includes general elements such as abstraction and decomposition in the problem-solving process. However, from a “hierarchical" perspective, artificial intelligence solutions involve higher-level computational thinking elements compared to traditional machine agent automation solutions. How to start from the general meaning of computational thinking and combine it with the unique attributes of artificial intelligence to clarify what level of computational thinking elements should be possessed in the current and future development process of artificial intelligence has become a problem that needs to be deeply explored.

3 Artificial intelligence requires computational thinking

Artificial intelligence is a data-driven computer science, whose core technology is the ability to express data and retrieve information through retrieval. Below, we will delve into the knowledge system in this field.

In the era of massive data, it is necessary to effectively solve the problem of data acquisition and as shown in Figure 1. Data based Computer Science System., simplify complex problem modeling by using various common knowledge expression techniques, such as propositional thinking, predicate thinking, inductive thinking, etc. Among them, due to the crucial role of data, techniques such as Bayesian theory, nonlinear regression, and fuzzy regression can also be used to handle complex situations, in order to better understand and solve data problems.

The problem of data-driven and data expression is also a complex process that can be achieved through different methods. The most common method is to analyze algorithm implementation and visualization. In addition, as shown in Figure 2. Data driven and data expression issues. modern intelligent algorithms such as local optimization, annealing algorithm, and particle swarm optimization can also be used to solve problems. Moreover, digital twin technology, virtualization technology, and cloud computing technology can be utilized to help solve this problem.

As shown in Figure 3. Data feedback method. Data feedback can be obtained through the application of machine learning, neural networks, and the latest models to study cutting-edge technologies, providing positive impetus in the era of data explosion. At the same time, exponential calculation optimization can greatly improve computational efficiency and decision-making accuracy, thereby achieving fast and accurate prediction of complex problems.

The six elements of computational thinking include:

• Decomposition: The problem is decomposed layer by layer, and the scale of the problem continues to shrink until it can be solved;
• Abstraction: Identify patterns from a wide range of instances and construct problem-solving models;
• Algorithm: The solving steps of a problem are algorithms, which require massive data support;
• Debugging: Tracking the process of the method, limiting the scope of the problem, identifying the root cause of the problem, and finally solving the problem requires a large number of sample and test sets;
• Iteration: Don't expect to solve problems at once, gradually strive for excellence;
• Generalization: Extending patterns to different specific scenarios for validation.

In summary, the manifestation of artificial intelligence in various industries cannot be separated from data, and data cannot be separated from the foundation of computational thinking. Computational thinking can also support various algorithm models, thereby helping to improve the value of artificial intelligence. This cycle constitutes a series of artificial intelligence foundations such as intelligent modeling, formal constraints, abstraction, disassembly, and merging of computational thinking.

4 Cultivation of Computational Thinking in Artificial Intelligence Teaching

4.1 Course Overview

The following text takes "Computational Thinking and Computer Science" as an example to elaborate. This course is a general education course aimed at all undergraduate students, aiming to popularize knowledge of computer science and artificial intelligence, and enhance students' innovation and practical abilities. Through AI assisted fusion learning of computer science knowledge, students can master the ability to design simple algorithms using computational thinking, laying a solid foundation for subsequent computer courses and artificial intelligence courses.

The characteristics of this course construction are reflected in: AI assisted - keeping up with the times; Theory and Practice - Case Study Integration.

Innovative teaching methods and their effectiveness are reflected in: establishing and improving a high-quality teaching resource system; Establishing a discipline competition training system and significantly improving academic performance; Establish an AI assisted resource library.

In today's era of rapid technological development, artificial intelligence education has gradually become an important component of the education field. In artificial intelligence teaching, the cultivation of computational thinking is particularly crucial. It not only helps students better understand and apply artificial intelligence technology, but also lays a solid foundation for their future innovation and development in the digital society.

4.2 Training Program for Computational Thinking

Computational thinking is a way of thinking that enables problem decomposition, abstraction, pattern recognition, algorithm design, and evaluation of solutions. This article proposes a practical plan for cultivating computational thinking for applied science and engineering talents under the background of AI general education (as shown in Figure 4. AI Fusion Teaching cultivates Computational Thinking), which integrates five “AI content+tools" into the design and implementation of the five teaching stages of the course.

• Problem decomposition: By selecting hot events and cases that are relevant to the teaching content and can be understood by students with different foundations, while attracting students' attention and interest, analyzing events, problems, establishing models, and attempting to solve them.
• Abstract AI Assistance: Clarify problem outcomes, provide clear goal directions to students with different foundations, use AI tools such as DeepSeek, Kimi, etc., and use different instructions and prompts to obtain different solutions.
• Algorithm learning: Combining basic programming theory, analyzing Al proposed solutions, analyzing underlying logic, and vertical deep learning.
• Algorithm design: Through classroom learning, LeetCode forum communication, and other methods, students are encouraged to actively learn and deeply participate in pseudocode construction in class;
• algorithm evaluation. Through self built evaluation web-sites and online compilation by LeetCode, competitions, and other means, students' learning effectiveness is tested, the advantages and disadvantages of different algorithms are understood, and their ability to apply what they have learned is evaluated.

4.3 Cultivation Practice

4.3.1 Al Assisted Keeping Up with the Times. This course is aimed at students with different foundations, using different case studies and AI assisted tools of different categories to complete problem decomposition and abstraction. Taking image data classification as an example, from the perspective of computational thinking:

• Problem decomposition: Image data needs to undergo digital image recognition processing and image category classification.
• Abstract AI Assistance: The image classification problem is abstracted into digital feature extraction and multi classification mod-els. The image dataset is divided into a sample set and a test set, and DeepSeek is used to determine the intelligent classification algorithm - Convolutional Neural Network (AlexNet).
• Algorithm learning: Analyze the implementation code of the Convolutional Neural Network (AlexNet) proposed by AI tool TongYi Qianwen and WenYan YiXin, including input layer, convolutional layer, activation function, pooling layer, fully connected layer, output layer, loss function, backpropagation, iterative training, an-alyze the implementation principle of extracting and processing image data using Convolutional Neural Network, draw analogies, and think about the computational thinking logic and problem-solving process in code writing.
• Algorithm design and evaluation: Referring to the AlexNet implementation code, comparing the pain points of actual problems, considering the correlation between backpropagation and data feedback, and combining the iterative thinking in computational thinking, attempting to design pseudo code that conforms to the existing image dataset classification, and completing the classifi-cation algorithm operation and quality judgment on the self built evaluation system (xujcoj.com) and LeetCode.

Summary: Reviewing the process of compiling algorithm code, understanding how the code solves image classification problems, comparing the differences between normal thinking, highlighting the problem decomposition and abstract modularization of computational thinking, and further clarifying the concept and practical application significance of computational thinking in students' cognition. Thus, artificial intelligence technology, including AI tools, neural network algorithm implementation, etc., plays a positive role in promoting the cultivation of computational thinking.

4.3.2 Practical Teaching Evaluation. As shown in Figure 5. Self built evaluation website, the self built evaluation website is divided into sections such as "Competition", "Homework", "Questions", etc., targeting different teaching environments and student groups. The website has been integrated into the teaching process, and the question bank is designed to include simple basic questions and competition level questions, with nearly 4000 questions. The running time of the system programming code can provide feedback on the effectiveness of students' computational thinking cultivation. In the later stage, a competition section can be added to further determine whether the foundation of computational thinking is stable.

5 Conclusion

Computational thinking is the core of practice in the field of artificial intelligence. With the development of artificial intelligence technology and the transformation of problem-solving approaches, computational thinking should present characteristics that are in line with artificial intelligence, thus reflecting differences in its elements that distinguish it from general information technology application fields. The development of artificial intelligence that meets the intelligence needs of real-world scenarios cannot be separated from the application of computational thinking. By designing corresponding scenarios, learners can continuously improve and further develop their computational thinking literacy in solving real artificial intelligence problems, cultivate innovative talents in artificial intelligence to move towards the realization of general artificial intelligence, and unleash the value of serving the future.

Acknowledgments This work is supported by the Fujian Province Undergraduate Education and Teaching Research Project(FBJY20240199) and Xiamen University Tan Kah Kee College Campus level "Cultivation Project" Incubation Project (PY2024L01). The authors would like to thank the reviewers for their valuable suggestions and comments.