Week 1

TOGETHER IN ELECTRIC DREAMS – Blog 1

Week 2

TOGETHER IN ELECTRIC DREAMS – Blog 2

Week 3

TOGETHER IN ELECTRIC DREAMS – Blog 3

Week Three Reflection

The focus of this week’s was on sound, music, and voice. In class, we practised and learned various sound-generating AI tools. After class, I also explored AI-generated music works. Among them, “Heart on My Sleeve” by TikTok user ghostwriter977 caught my interest. This song utilises AI-generated voices imitating Drake and The Weeknd while incorporating popular lyrics from celebrities like Selena Gomez and Justin Bieber. From this song, it can be observed that generative AI could easily copy the voices of others, blurring the lines between human artistry and AI-generated content. I believe that the existence of this type of music is controversial in terms of ethics and legality. While such songs may indeed garner popularity in the short term, they have implications for the artists themselves, record companies, copyright issues, and the music industry as a whole. However, this song has now been taken down due to numerous threats

Based on the research mentioned above, I created a music video for this week’s sketch. I used AI generative software to extract Selena Gomez’s voice and sing Justin Bieber’s “Love Yourself.” Overall, I believe the effect is quite similar, with about 70% of Selena’s voice. In creating this song, I wanted to understand the difficulties and processes behind it. If a beginner can produce such music in a short time, similar actions in the future may cause more controversy. Additionally, I also made a video for this song. I recorded their first collaboration stage and converted the video into a pixelated style, giving the entire MV a humorous, funny tone with a hint of nostalgia and reminiscence. I believe using AI generative tools in this way is inappropriate. If it’s simply for humour or paying tribute, it might be more suitable. If the purpose is profit-driven, there may be legal implications.

AI technology indeed assists musicians by providing creative inspiration, expediting the music production process, and utilizing deep learning systems to create various music styles and performance forms, thereby pushing the boundaries of music creation further. Additionally, AI can generate personalized content based on user preferences, offering customized music experiences tailored to individual users and market demands, which is currently beyond human capabilities. If used in this manner, generative AI can provide more convenience and insights to the music industry. However, AI has always been a double-edged sword. How it is used determines what people can gain, whether it be enhanced creativity or potential risks.

References

Week Two Media Sketch

Week Two Vlog Reflection

Week One Reflection

Recently, the media work that has captured my interest is “The Next Rembrandt” by J. Walter Thompson Amsterdam and Microsoft.
This project involves the use of analytical algorithms and a study of Rembrandt’s works to generate paintings mimicking his distinctive style. What fascinates me about this piece is the concept of “resurrection.” The emergence of AI enables the revival of elements that were once lost. Such works not only pay homage to historical artists but also capture the essence of the artist to create new pieces. Last week, I came across a video created using AI on a forum. In a satirical manner, the creator “resurrected” Empress Dowager Cixi, transforming an old photograph into a dynamic form. Although virtual, the ability to revisit history in such a manner is both intriguing and entertaining.

This week, we utilized AI to create our own media sketches, and overall, I found the experience quite interesting. However, there exists a disparity between the final output and our expectations. We used Runway as the tool, and I’m uncertain whether the perceived gap is due to the tool still being in its early stages of development. While generating videos using text, I observed that AI exhibits a certain degree of accuracy in recognizing and generating content but lacks precision. In comparing emotions and actions, AI seems to be more proficient in capturing actions than emotions. In addition, I incorporated real images, hoping for a secondary generation based on these pictures, but the results were not satisfactory.
I believe AI places high demands on the clarity and conciseness of instructions; our provided instructions need to be clearer and more concise. Although the text box indicates a maximum of 320 words, I feel that AI struggles to generate content responsibly in a single attempt. This time, our team controlled the video duration at 4 seconds each. Next time, we could adjust to a longer duration to see if it conveys better results.

The reading that captured my attention in this week’s reading is the distinction between generative artificial intelligence (AI) and other forms of AI. Generative AI creates new works by gathering information. The reading mentions that “Generative AI is an automated form, and its outputs in this form are narrow and limited. These systems do not comprehend what they generate, nor can they evaluate their outputs, sometimes leading to misunderstandings.”(Garon 2023)
Reflecting on our media sketch this week, AI did indeed generate something resembling the overall direction but exhibited issues and lacked emotion. Currently, most AI is business-oriented, streamlining complex work, automating routine tasks, and increasing productivity and profitability. However, generative AI produces content rather than regular work. In the media world, where precision in content and final presentation is crucial, generative AI needs accurate capturing abilities. Similar to “The Next Rembrandt,” if AI cannot precisely capture the essence/artistic habits/core topic of the artist, it becomes challenging to create satisfying new works. Reading this article informed me about the distinctions in various forms of AI, and I look forward to trying more AI tools, using generative AI to create works in different media forms.

References