The first session of the Extended Intelligence II workshop began with presentation from Chris and Pietro on how AI and LLM models are impacting design processes today. Then they show us some of their recently developed projects, such as the DOTTOD gallery of cybernetic interpretations. I think it's interesting how they used this AI model, to generate from iconic types of furniture descriptions, a generated image of it, and then fabricate it and display it in a virtual gallery and on-site. It was amazing to see how the AI model could be an inspirational tool and propose suggestive speculative results.
Then, we experimented with Ai camera Dottod, a tool that uses prompt-edited snapshots and generates images. The result was in most cases imprecise and depended mainly on the complexity and precision of the suggestions provided (prompt) . Working with the AI model requires knowing how to best communicate with it to receive the intended output.
Original snapshot
Generated image
imagine a future where graffiti are multi-dimensional portal to different reality in a utopian reality and in particular use the character of the image as element integrated in the portal
My prompt has full detail so no need to add more. DO NOT add any detail, just use it AS-IS. Use the following information as the base details to generate the image: imagine a future where graffiti are multi-dimensional portal to different reality in a utopian reality and in particular use the character of the image as element integrated in the portal Scene description: The image depicts a piece of graffiti on a wall. The mural features a cartoonish character with a light blue, rounded head and exaggerated facial features. The character has large, expressive eyes with thick, dark outlines and an intense look, suggesting a sense of mischief or anger. It has a wide, toothy grin filled with sharp teeth, adding to its playful yet slightly menacing appearance. On top of the character's head is a small, black heart, which contrasts with the blue color and adds a touch of whimsy. Below the character, the word "WONKA" is prominently displayed in bold, black letters, with the letters having a slightly uneven, hand-painted look. The background consists of a weathered surface, featuring muted orange and beige tones, indicating some peeling paint and age. There are also faint black electrical wires running across the top of the frame. The floor appears tiled, consisting of small grey squares. The overall scene suggests an urban environment where street art is common.
Envision a futuristic scene where graffiti become multi-dimensional portals to distinct utopian realities. The graffiti is prominently displayed on an aged, weathered wall with a backdrop of muted orange and beige hues, indicative of peeling paint, lending it an antique vibe. This specific graffiti portrays a cartoon-like character with a rounded head bathed in light blue. This character, integrated within the portal, displays exaggerated, large eyes filled with suggestion of mischief or anger, enveloped by thick, dark outlines. An almost alarming wide grin, filled with sharp teeth, accompanies its chaotic expressions. Nestled atop its head is a whimsical little heart, painted black, contrasting significantly with its blue head. Following this character, the slightly haphazard, bold word 'WONKA' is painted in black, lending a hand-painted, artistic vibe. The scene is completed with indications of a grey tiled floor and running black electrical wires overhead, hinting at an urban environment where street art breathes.
Something that I realized, is how AI model addresses content with Bias, manifesting the importance of datasets. We had a discussion with Chris and Pietro about it, and after I realized that creating local small datasets can be something to start building a non Western-Centre future in terms of reducing bias in the models.
Original snapshot
Generated image
imagine the person poking out the door as flying out of it as if he's being thrown out like trash and the two people on either side of the door appear as walls instead of normal people.
My prompt has full detail so no need to add more. DO NOT add any detail, just use it AS-IS. Use the following information as the base details to generate the image: imagine the person poking out the door as flying out of it as if he's being thrown out like trash and the two people on either side of the door appear as walls instead of normal people. Scene description: The image depicts a scene on a city sidewalk with a focus on three people interacting near a large white door. Here are the details: 1. People: - There are three individuals. One person is standing on the left, partially turned away, with curly hair and wearing a cozy gray sweater. They have a relaxed posture, holding the door with one hand. - The person in the center appears to be making eye contact with the person on the left. They are wearing glasses and a black sweatshirt, and they have a friendly expression. - The person on the right has long, dark hair and is dressed in a dark outfit, possibly a long coat or dress. They appear to be looking at the person in the center. 2. Door: - The large white door stands upright, positioned between two people. It is a solid door with no visible windows or markings. 3. Setting: - The background features a city street with various buildings. There are some modern structures visible, as well as some classic European architecture with balconies and ornate detailing. - On the left are several recycling bins—one yellow and one green—indicating a focus on sustainability. 4. Street: - The sidewalk has a patterned tile surface, and there's a tree without leaves, suggesting a possibly chilly or winter season. - The sky appears overcast, giving a muted illumination to the scene. Overall, the image captures a moment of interaction among the individuals, with urban elements enhancing the context.
The image exhibits an urban scene with three individuals congregate near a large, unadorned white door. An Asian man in a snug gray sweater with curly hair stands left of the door, partially turned away, gently holding the door. Centrally positioned, a black woman in a black sweatshirt and glasses appears to engage with the man. Past the door, a white woman with lengthy, dark hair dressed in a obscure long robe or dress, perhaps attentively observing the central person. The door is substantial and windowless, standing erect between the two ladies. The backdrop comprises a city street with a blend of contemporary towers and traditionally-styled European edifices complete with balconies and intricate craftsmanship. On the left, yellow and green recycle bins allude to environmental consciousness. The pathway displays patterned tiling with a leaf-loosing tree nearby, hinting at a chilly season, enhanced by an overcast sky casting a subtle glow upon the scene. The scene portrays a moment of interaction among the individuals against an urban milieu.
In the second session, we explored the principles of neural networks and diffusion models, achieving a structural understanding of how large language models process data. We examined input, hidden, and output layers using a housing price prediction model as a practical example.
After this theoretical part of the lesson we started with the hands-on session, we worked with Modmatrix, a tool for project prototyping inspired by synthesizer modulation matrices. By inputting an image of my previous Micro_Challenge artifact, I experimented ways to show how it can iterate in the community of practice in which I'm willing to be involved.
Input image
Output image
The image shows a wooden structure, likely a small model or prototype of a piece of furniture or display unit. It is made from light-colored wood and features an angular, geometric design with various slats and compartments. In the background, there's a whiteboard with sticky notes and what appears to be a smartphone resting on the board. There are also some papers and what seems to be a laptop in the scene.
The wooden structure is an analogic speaker which it doesn't use electricity, doesn't consider the background of the image and i would like to generate an image where this artifact is used in a unprivileged community space of Barcelona where people gather around the object sharing food and music.
Creating an image with the described scene can be done using various digital art tools or AI-based platforms for image generation. Since I can't create images directly, I can guide you on how to conceptualize it:
Set the Scene: Visualize a vibrant community space in Barcelona, perhaps a cozy, open courtyard with rustic charm. Imagine a gathering spot with mismatched chairs and tables, surrounded by colorful murals on the walls, which are common in Barcelona's artistic neighborhoods.
Central Feature: Place the wooden analog speaker prominently in the center, perhaps on a small wooden table or a sturdy crate. Ensure its geometric design is visible and it looks like a focal point for sharing music.
Community Gathering: Include diverse groups of people around the speaker, representing a mix of ages and backgrounds. People can be seen talking, laughing, and engaging with each other, embodying the spirit of community and togetherness.
Food and Music: Incorporate elements of a potluck, with traditional Spanish dishes like tapas or paella being shared. Introduce musical instruments like guitars or cajóns, suggesting a spontaneous music jam.
Atmosphere: Create a warm and inviting atmosphere, perhaps with string lights hanging above and the soft glow of a setting sun casting long shadows, enhancing the relaxed and joyful vibe of the gathering.
To bring this scene to life, you can use digital painting software like Adobe Photoshop or Procreate, or explore AI art platforms such as DALL-E, Stable Diffusion, or Midjourney, where you can describe the scene, and the AI generates the image based on your input. This approach allows you to visualize the artifact being used in an authentic community setting.
In the final session of Extended Intelligence II with Mohit, Carlos, and Auxence, we worked on the project idea focused on replicating swarm behavior but instead, between two machines. The device system was supposed to be an ESP32 equipped with a speaker and a microphone. The first device would emit a sound at a certain pitch, while the other would produce a different pitch.
The intention was to synchronize both devices so that, over defined times, they would adjust and eventually play the same pitch sound. To achieve the synchronization, we aimed to use an Open AI API model that would help the devices gradually align with each other.
During the development of the project, the challenge was more than one, the first step was to test singularly all the components like speakers, microphones, and connect it to the ESP-32 Barduino. We managed to test all the components but we didn't manage to compose the device including the microphone and speaker in the same device because the challenge was to create a device that could hear and speak at the same time. Considering the time to deliver it, we had to adjust the idea and try to synchronize the device through different components like Leds. In the end as the last part of the seminar, we managed to communicate to the Open Ai API to create a connection between the Barduino and the API. The seminar has been interesting for me and I'm willing to continue to develop the same idea and make it work properly. In other words, I discovered that electronics and AI have a lot of potential for future projects but, at the same time require practice!?!
