Mikaela's San Giorgio – Digital Portfolio

Digital Artefact Exploration Theme: Exploring AI Media

*AI tools were used to enhance the readability of this blog post and to ensure it aligns with the marking criteria.

In the first phase of BCM 114, I focused on exploring AI media, particularly wanting to experiment with LLM’s I had not used before. My initial goal was to observe what AI characters were already out there and then explore how they were created with AI. I did not expect to end up with a virtual human I actually liked just yet – but I did! I naturally progressed to prototyping using OpenArt and I was happy with what unfolded. Using the ‘Design Thinking Process’ from BCM 114 – Module 2, I observed, framed, ideated, prototyped, and tested, cycling through all five steps to develop a virtual human called ‘Kumere’.

My specific interest was creating a virtual human or AI influencer. I wanted to understand why virtual humans / AI influencers like Aitana (who has over 300k follows on instagram) are so successful. I now think it is largely because of how consistent they look across posts, which allows audiences to follow an authentic story and connect emotionally. While I have explored ChatGPT extensively, I had little experience with other AI programs, so this project became an opportunity to broaden my knowledge of generative tools. I started by researching online tutorials and found a youtube video recommending OpenArt to make a character like Aitana.

Encouraged, I explored OpenArt’s video-making feature. In my first attempt at creating a video of Kumere, I prompted her to wear a white bikini. The result? She was in a bikini… but still wearing part of her jeans! (a hilariously imperfect outcome). Next, I asked for an ocean background and for her to be sitting on a boat fishing, but the AI accidentally generated two versions of her and kept the original background, making the scene look super funny.

This led me to think, maybe using one AI program I am familiar with could help guide the one I am new to using. I experimented with ChatGPT to generate more sophisticated prompts for OpenArt, hoping the LLMs could “speak the same language” more effectively than I could on my own. I got this idea from some conversations in the #exploring-ai-media channel on the discord page. Outputs improved slightly, though inconsistencies remained. Cycling through the ‘Design Thinking Process’ repeatedly taught me how crucial iteration, experimentation, and prompting are, and how unexpected outcomes can spark learning.

I have circled through the ‘Design Thinking Process’ many times over the last few weeks. Observation gave me ideas for what AI I should learn, and framing helped me select the angle of approach for Kumere, deciding which aspects of her appearance and personality to emphasise. Ideation helped me to stay on track, by keeping my options open to see the bigger picture. Prototyping and testing taught me the most – just give it a go! Each cycle revealed insights I would not have noticed through observation alone, helping me refine prompts, outputs, and Kumere’s personality. The process built my confidence experimenting creatively with AI media while thinking about how consistent appearance supports storytelling and audience connection. The next step I want to focus on is doing testing to reduce guesswork and produce more reliable outputs with what Kumere does in her videos.

Recently I connected with a group of BCM 114 students on Zoom to share our thoughts on the exploration theme of ‘Viral Media‘. Although not my main focus theme just yet, discussing it helped me see new possibilities for Kumere and AI-generated characters, and it was a valuable space to iterate and learn from peers’ creative journeys.

So far in this DA I’ve learnt that iteration, prompting, and experimentation are essential when working with AI tools. It takes time to get outcomes that you want – you often need to change tact. Using ChatGPT to guide OpenArt highlighted how maybe when done correctly, one LLM can help navigate another. Cycling through the five steps of the ‘design thinking process’ accelerated my learning, confidence, and consistency in creating Kumere. Next, I plan to develop Kumere’s personality and bring her onto social media!

Stay tuned….

REFERENCES

ABC listen, 2014, Going viral- the 19th century way, viewed 18 August 2025 https://www.abc.net.au/listen/programs/futuretense/going-viral–the-19th-century-way/5251686
ABC listen, 2025, Ai influencers infiltrating our social media feeds, ABC Radio National, Australia, viewed 20 August 2025 https://www.abc.net.au/listen/programs/lifematters/ai-influencers-infiltrating-our-social-media-feeds/105566248
Builtin, 2025, Builtin, viewed 19 August 2025 https://builtin.com/articles/ai-influencer
Cognitive Creators, 2020, What is a prototype?, online video, viewed 19 August 2025 https://www.researchgate.net/publication/209409849_Framing_Toward_Clarification_of_A_Fractured_Paradigm
Dam, R F and Siang, T Y 2020, Interaction design foundation, viewed 20 August 2025 https://www.interaction-design.org/literature/article/introduction-to-the-essential-ideation-techniques-which-are-the-heart-of-design-thinking
Discord, 2025, Discord, viewed 10nd August 2025 https://discord.com/
Easy Marketing, 2023, Product concept testing, online video, viewed 21 August 2025 https://www.youtube.com/watch?v=NSYr9PI7VBQ
Entman, RM 1993, Framing: toward clarification of fractured paradigm, Recharge Gate, Northwestern University, viewed 19 August 2025 https://www.researchgate.net/publication/209409849_Framing_Toward_Clarification_of_A_Fractured_Paradigm
Google for Developers, 2023, What are large language models (LLMs)?, online video, Goodle for Developers, America, viewed on 20 August 2025 https://builtin.com/articles/ai-influencer
The Clueless Aigency S.L, 2025, The Clueless Agency, viewed 20 August 2025 https://www.theclueless.ai/project/aitana-lopez
Virtual Humans, 2025, Virtual Humans, viewed 20 August 2025 https://www.virtualhumans.org/#influencers
Zoom, 2025, Zoom, viewed 18 August 2025 https://www.zoom.com/