Integrating Adobe Firefly in the Classroom

Before asking students to use Adobe Firefly, educators should first consider experimenting with Adobe Firefly themselves. Experimenting with Adobe Firefly can offer insights into the capabilities and limitations of the tool. After learning the tool, consider modeling or incorporating it into a lesson that demonstrates its applications to guide students on how to make the most of Adobe Firefly. For example, instructors can emphasize the importance of providing context and specificity in their prompts to achieve the desired outcome. Students should also be encouraged to experiment with different prompts to discover the most efficient ways to achieve their desired results. 

When incorporating tools like Adobe Firefly into the curriculum, faculty should provide students with clear instructions on the tool's technical aspects as well as the ethical considerations surrounding AI-generated content. One approach to promoting ethical AI integration is by engaging students in the co-creation of guidance through community agreements on AI use. This collaborative effort fosters a sense of shared responsibility and accountability among students. Faculty, in partnership with their students, can develop guidance that addresses both the potential and limitations of AI technology, and lay the foundation for responsible use. 

Another aspect to consider when integrating generative AI tools like Adobe Firefly is whether students will be required to acknowledge the use of AI-generated content. Faculty should provide clear guidance to students on how to cite AI-generated content accurately for the course or utilize the most recent updates in their respective style books such as APA, Chicago, and MLA. 

Faculty should consider the advantages of pairing AI-generated image assignments with reflective writing. This combination not only fosters critical thinking but also empowers students to articulate the rationale behind their creative decisions. For example, students can reflect on the choices they made while crafting images in Adobe Firefly. The reflective process can also prompt students to consider why they selected specific elements, effects, colors, or tones in their creations.

Visual representations may not offer equal accessibility to all learners, particularly to those with visual impairments or individuals who lack familiarity with the specific graphic format in use (CAST, 2018). When designing generative AI assignments, it's important to consider providing alternative assignments that enable students to show they've met the learning objectives using alternative formats. Alternative assignments can include non-visual alternatives such as text or spoken descriptions.

Prior to deciding if and how to integrate generative AI technology, it's important to reflect on your course’s learning goals. Does the assignment that utilizes generative AI technology align with an existing goal? Will the use of generative AI technology facilitate the development of new skills and knowledge that are relevant to the course's learning goals? How might the inclusion of generative AI technology expand students' understanding of the subject matter or field of study?

As with any AI-powered tool, Adobe Firefly's output is influenced by the data it is trained on. Faculty and students should be aware that the tool's content generation may reflect the biases inherent in the data sources. This awareness is crucial to ensure that the educational content produced with Firefly is inclusive and unbiased. To foster student understanding, faculty may want to facilitate a discussion about the tool's potential biases that can encourage students to critically evaluate the content they encounter as well as deepen their understanding of the complexities surrounding AI. 

While Adobe Firefly offers impressive capabilities, it's important to acknowledge that there may be instances where the generated images include body features that appear unrealistic. This is particularly relevant when designing content that involves human representation. With the latest version, Firefly Image 2, the images rendered are of higher quality including improved human body features.

Unrealistic body features using Firefly Image 1 (left) and Firefly Image 2 (right).

Prompt: Professors meeting with the dean discussing AI virtually with eager smiles

Additionally, users should be aware that text may sometimes appear distorted and unclear in the images generated by Firefly. This limitation may affect the legibility of textual content within the images. Faculty and students should exercise caution when relying on Firefly for text-based visual materials.

Distorted text using Firefly Image 1 (left) and Firefly Image 2 (right). 

Prompt: Butterfly flying by the Miami River you can see a sign that states Miami River

Adobe Firefly is trained on data primarily sourced from Adobe Stock. This may limit the range of images it can generate, potentially impacting the tool's versatility in certain educational scenarios. Given Firefly's training on Adobe Stock, students encounter limitations when attempting to generate images resembling those of well-known artists. For instance, they cannot directly prompt Firefly to replicate an image of a cell in the style of Van Gogh's 'Starry Night'. Instead, they would need to adapt their prompts and utilize various effects, colors, and tones to craft something inspired by 'Starry Night.' As a result, educators should be aware that the tool's image database is not exhaustive, and they may need to complement Firefly-generated content with external resources where necessary to meet specific learning objectives.

Prompt: The cell with visible brushstrokes rendered in a bright opulent color palette with wavy patterns

Style: Painting, Impressionism