Week 3
Making of a Seeing Object
Just as a recap, for this artefact I wanted to create this effect that shows the idea of being watched, getting at the idea that your information is always being studied by AI LLMs while you’re talking to it. While it lets the AI personalise your experience, it also learns more about you, becoming more like you and sounding more human, except it’s really just mimicking your own behaviour back at you.
To achieve this, I moved away from chat interfaces and started using Arduino, along with an ultrasonic distance sensor to detect when someone is nearby. What I wanted to do was use an LCD screen to show a visual of an eye opening and closing depending on the distance detected. The closer someone is, the wider it opens, and the further away they are, the more it closes.
I went on a trip to Sim Lim Tower to look at the screen options available, and decided to go with a larger blue bitmapped LCD screen. I only bought one first so I could test it out, and if it worked, I could connect another screen later.
I went through a lot of tutorials online and tried the most straightforward wiring I could find, just to understand how to power the screen and what each of the pins were for.
In the end, I managed to get the screen up and running. I then changed the wiring so that the screen pins weren’t directly in the breadboard, but connected through jumper wires instead. This would make it easier later on to position the screen where I need it, rather than having it fixed to the breadboard.
The next thing I needed to figure out was how to display my own images on the screen. While doing research, I realised that this type of screen can’t just have images uploaded directly, they need to be converted into bitmaps beforehand.
What I did was take three separate images and run them through an image bitmap converter. These were then stored as three different arrays, one for each image.
The distance sensor detects how many centimetres something is away. Anything over 30cm shows the closed eye, around 15cm shows a half-open eye, and around 7cm shows the eye fully open.
Setting it up, it doesn’t work too badly. The only issue I have right now is that using three different states makes it feel a bit too static. For example, if someone stays at around 15cm for a while, the half-open eye just sits there, which feels a bit awkward. It gives off slightly lifeless vibes at the moment.
I also realised that using a set of bitmapped images is quite inefficient, which is probably why it feels so static. If I wanted more states, I would need more bitmap arrays, which could get messy quite quickly.
I think it would look a lot smoother if I coded the shapes to animate or translate instead of relying on static images (Claude to the rescue).
Another thing is the limitation of the screen resolution. I initially chose this blue screen because I thought it would really emphasise the “technology” aesthetic. But thinking about it now, with AI being experienced through high-resolution interfaces, it might make more sense to use a higher quality screen instead.