Increasing middle school girls’ critical engagement with AI through lightweight workshops · ICGS Research Library

Title

Increasing middle school girls’ critical engagement with AI through lightweight workshops

Author

Solyst, J., Axon, A., Stewart, A. B., Eslami, M., & Ogan, A. (2023)

Year Published

2023

Description

With the increasing presence of AI in society it is becoming vital that students understand how to use this technology safely and are aware of its potential benefits, uses, bias, and impact on ethics and privacy. AI is prevalent in everyday life, and girls are highly likely to be exposed to AI regardless of which post-school path they pursue. Researchers from Carnegie Mellon University in the USA have undertaken a project to gain a better understanding of middle school girls’ “perceptions and knowledge gaps about AI” (Solyst et al., 2023, p. 807). By developing a lightweight educational workshop (a workshop less than three hours long), the authors explored girls’ perceptions and approaches to AI with the aim of helping them avoid misinformation, provide them with useful learning materials and “lend insight into a more just future” (Solyst et al., 2023, p. 807).

Currently, learning opportunities that are designed “to address children’s knowledge gaps in AI literacy” largely focus on the technical aspects of the technology. This excludes many of the ethical questions associated with AI (Solyst et al., 2023, p. 807). The researchers who developed this paper were also concerned about the limited understanding “of how girls perceive and learn about AI”, with a risk that this may “potentially [compound] existing inequities in AI representation” (Solyst et al., 2023, p. 807). This project explored a learner-centred education format to support middle school girls to fill their knowledge gaps around AI in a shorter workshop form. This is an important new contribution for educators as it provides an alternative to existing programmes that are more than three hours long, or are run as an intensive workshop session.

The workshops were specifically targeted at middle school girls and assumed that participants had no prior computing or AI-related knowledge. They were focussed on “critical and creative thinking about AI systems and ethics, and supported learners in thinking about training data” (Solyst et al., 2023, p. 807). The workshops were also designed around an asset-based approach (rather than a deficit-based approach), and drew on girls’ prior “knowledge and interests as a base for further learning and engagement” (Solyst et al., 2023, p. 807). This is important because it has been shown to be a successful way to “support girls of diverse backgrounds in learning computing” (Solyst et al., 2023, p. 807).

The researchers considered girls’ understandings and perceptions before, during and after the workshop through observations of the workshop session, surveys and interviews with participants. The workshop was run as a standalone module offered within an all-girls computing camp. Participants for the project were recruited via BoltGirls (a robotics-focused organisation in the east coast of the US) and AmazingGirls (a general girls organisation in the southwest of the US). The workshop ran for approximately 90 minutes, which included educational content and interactive activities.

Workshop content was focussed on the concepts of bias, algorithms, and definitions of AI (including training data, power, and how AI can identify or misidentify content). During the interactive sections of the workshop, this was extended to consider how AI could be used to solve challenges in the girls’ own communities and a problem affecting the world more generally. This focussed on how AI could help solve the problem, but also risks of harm and how this could be mitigated. While the workshop was interactive, there was “less use of interactive tools” and a greater focus on “group discussions and ideation… as a means to understand opinions, perceptions, and knowledge gaps” (Solyst et al., 2023, p. 813).

Many girls who participated were able to describe AI without technical details prior to completing the workshop. There was an improvement in technical awareness following participation, and a “more complex understanding of AI” compared to the pre-workshops surveys (Solyst et al., 2023, p. 810). The girls also showed an increase in ability to talk about AI, with a much more nuanced approach to intelligence and AI, and comparisons between artificial and human intelligence. This included the recognition of different types of knowledge and intelligence, understanding the role of “informational knowledge and facts”, and being able to recognise limitations in the capabilities of AI technology (Solyst et al., 2023, p. 810).

The workshops discussed algorithmic bias, and girls were able to identify bias in the context of AI providing misinformation. This included concepts such as racism, and the ability to critically analyse content to discuss implications for society. Girls raised concerns about how AI uses data, especially in relation to privacy, although they did not always specifically link this to AI, and instead focussed on the risk of hacking when discussing chatbots and digital assistants. While some participants were able to identify t