人工智慧素養問卷開發與驗證

本研究旨在開發一個測量大學生「人工智慧素養量表(Artificial Intelligence literacy, AIL)」的量表。在全面推廣人工智慧知能的近代，試問教育工作者應如何判斷何樣內容應包含在AI課程中、並課程應如何設計，才能適用在不同背景之學習者身上？而最終在這些全面開展的AI通識課程中，我們所預期能夠提升學習者之人工智慧的「素養」為何？為解答這些問題，研究初期，根據相關文獻之探討及實際參與多樣的基礎人工智慧課程及工作坊後，研究者統整出三個面向包含：人工智慧知覺(AI Perception, AIP)、人工智慧適應（AI Adaption, AIA）、跨領域合作的態度(AI Teamwork Attitude, ATA)。接續，為進一步確認面向，本研究採問卷調查法，針對北部某大學選修通識課程「自然科學導論」的學生進行問卷調查，共收回845份有效問卷。本量表藉由探索性因素分析來確立問卷的面向，並進一步對大學生的人工智慧素養現況與背景資料作探討分析，分別進行獨立T檢定及單因子變異數分析(ANOVA)。研究結果顯示：(一)AIL量表之面向依序為：人工智慧知覺(AIP)、人工智慧適應性（AIA）、團隊合作知覺（AI Teamwork Perception, ATP）及團隊自我效能（AI Teamwork Efficacy, ATE）。(二)AIL量表之信度Cronbach’s α值為0.89，各面向Cronbach’s α值分別為：AIP 0.66、AIA 0.88、ATP 0.86及ATE 0.77，具有良好之內部一致性，總題數為23題。 (三)受試者背景資料中，男生在人工智慧知覺(AIP)、人工智慧適應性（AIA）及人工智慧素養(AIL)總分之表現顯著高於女生。(四) 受試者背景資料中，設計學院在團隊合作知覺(ATP)及團隊自我效能(ATE)之表現顯著低於其他學院；電機資訊學院在人工智慧適應性（AIA）之表現有顯著高於其它學院，人文教育學院則顯著低於其他學院。(五) 受試者背景資料中，7歲以下擁有3C在團隊自我效能(ATE)、人工智慧適應性（AIA）及人工智慧素養(AIL)總分之表現顯著高於其他年紀擁有3C。在各級學校中，大規模開設AI相關之通識課程，而未思考何為AI學習之目的，實為本研究之設計與施測背景。研究者根據分析結果指出AI相關之素養的提升應在上述四個面向（人工智慧知覺、人工智慧適應性、團隊合作知覺與團隊自我效能），並以此四面向作為短期內AI通識課程設計之參考。

關鍵字

人工智慧素養

並列摘要

The purpose of this study was to develop an instrument to measure college students’ artificial intelligence literacy (AIL). How could educators determine what kind of content should be included in the AI course, and how the course should be designed to be applicable to learners with different backgrounds? And eventually, how do we improve the "literacy" of learners' artificial intelligence in AI courses? Therefore, we established a questionnaire that can evaluate the artificial intelligence literacy of college students so that educators could consult it as a reference when designing AI courses in the future. We searched the literature and participated in artificial intelligence courses and workshops in order to define the abilities of artificial intelligence literacy. From these abilities, we came across three dimensions which include AI perception, AI adaption, AI teamwork attitude. And then, we used survey as a method, and collected 845 questionnaires from the college students who took the course in "Introduction to Natural Science". The dimensions in the AIL scale was developed with the result of our exploratory factor analysis. T-test and ANOVA were taken to analyze the relationship between the results of college students’ attitude to AIL and the background of the volunteers. The results show that: 1. The dimensions of AIL scale were as follows: AI perception (AIP), AI teamwork perception (ATP), AI teamwork efficacy (ATE), AI adaption (AIA). 2. The reliability were reported using Cronbach’s α as follows: Overall AI literacy was 0.89, and each dimension was reported as AIP 0.66, AIA 0.88, ATP 0.86, ATE 0.77. 3. Males had significantly higher performance than females in AI perception, AI adaption and overall AI literacy. 4. Students from College of Design revealed significantly lower level of AI teamwork perception and AI teamwork efficacy; students from College of Electrical Engineering and Computer Science revealed significantly higher level of AI adaption, whereas students from College of Humanities and Education were on the opposite. 5. For those who first had electronic devices under 7 years old, revealed significantly higher level of performance than those who first had electronic devices in other ages in AI teamwork perception and AI adaption. According to the findings of this study, an instrument for evaluating artificial intelligence literacy are presented. Further, some suggestions and educational implements are discussed for the design of current AI courses.