Skip to content
#AI

Leveraging Computational Thinking in the Era of Generative AI (AI column)

#AI #ML #Software #Architecture #Future

Interesting. article Computational thinking in the LLM era. After I became a member of ACM (Association of Computing Machinery) I began to receive collections of news, some of which I even manage to read. Specifically, this article was about computational thinking, a term that was popular in the 2000s and how relevant it is now.

Going back to the definition, computational thinking is shaped in terms of the questions, “What can humans do better than computers?” What can computers do better than humans? (Wing, 2006). In fact, in the 2000s, there was a flood of articles in which authors tried to determine what cognitive and social skills a person needs for problem-solving processes in general and for solving computational problems in particular. Today, the key components of computational thinking are considered (Wing, 2010)

  • Formulation of the problem
  • Decomposition of the problem Organization and logical analysis of data Presentation of data using models and simulations
  • Abstractions. Proposal and evaluation of several solutions to the problem
  • Implementation of the chosen solution
  • Generalization. These computational thinking skills are now recognized as important not only in the context of computer science, but also as important skills for everyone in the world. 21 century and can be applied in different contexts (Günbatar, 2019; Wing, 2017).

Further, the authors of the article note that in the era of LLMs, prompt engineering is actively developing, which is defined as such. (Dang et al., 2022)

Systematic practice of constructing prompts to improve the generated output of a generative model This is a new area of research that asks what best practices exist to form prompts for GenAI tasks in general and for programming tasks in particular. Funny enough, this new field of research could quickly be old after the emergence of new LLMs that do prompt engineering within themselves:)

But the authors note that computational thinking is needed for effective prompt engineering, which in turn allows you to pump computational thinking, which leads to a positive feedback loop. Interestingly, Yael Erez, co-author of this article, teaches an Introduction to Computer Science course. (CS1) undergraduate students in the specialty "Computer Science". In the fall semester. 2023 During the year, students were assigned several tasks, including solving development problems using GenAI applications such as ChatGPT. And the data on this training confirms the above link between computational thinking and prompt engineering. Computational thinking is also useful in the era of GenAI. And while GenAI itself can make it easier to solve a programming problem even for complex systems, thereby eliminating the need to master a particular language syntax, they still lack computational thinking. Therefore, computational thinking of people is necessary in the era of GenAI.

P.S. It is interesting to look at the latest achievements of foundational models, because the situation is developing very rapidly. And while computational thinking of people still helps LLMs, then everything can change:)

#AI #ML #Software #Architecture #Future