Err
Unmatched perceptions
Client
Academic
Year
2021
Category
Speculative Design
Type of Work
Design Research,
Product Design
Duration
5 Days
"In the kaleidoscope of perspectives, mismatched perceptions create the intricate beauty of understanding."
Err - is a concept to visualize the unmatched perceptions generated by an Al system in the form of an interactive installation.
It was fascinating to know and discuss beyond the technical side of Al, the moral and ethical aspects of it as part of the Open Elective 'Artificial Stupidity and Machine Unlearning!'
Being introduced to creative coding and how Al as a component can aid it- is another key takeaway from the course that helped the ideation of Err.
Err will be looking forward to raising questions and evoke thoughts on, Having unmatched perceptions and a way forward, human vs machine, Black-box algorithms and influence of Al-made decisions in our life and Misinterpretation of information and forced perceptions.
Err hosts an Al model to receive a caption as text input from the viewer. The received caption will be visualized with the help of the AttnGAN model and will be presented to the viewer.
And if the outcomes do not match, the process goes on to visualize the interesting conflict between the models. The result of each iteration will be displayed to the viewer to analyze and question.
Speculative design is an effective tool to question the future and influence current actions and their reflection on the future. Artificial intelligence as a tool hence needs to be understood and questioned before it is part of our conscious
A simple scenario representing the bigger picture provoked a lot of questions towards the credibility of such crucial decisions being taken by the Machine Brains rather than humans. It is also understood that machines can also have conflicts, and the resolution can compromise and shift from the reality of truth.
The Al systems generally are understood as smart robust systems capable of making foolproof decisions. On a closer look, the decisions are purely depending on the data it trained with. This adds to the ambiguity of how to trust any such systems if the data fed was not right and the decision is flawed. Hence from a point of view to depending on Al systems, the shift happened to question them and their decisions for possible errors, compromises, and biases.
Prototype Documentary
Learning
Err was a part of an Open elective course at NID. In Pawel Pokutycki's open elective course, "Artificial Stupidity and Machine Unlearning," I immersed myself in a transformative exploration of design ethics and humane tech. Guided by Pokutycki's emphasis on non-commercial values, the course propelled me beyond traditional boundaries, fostering skills in speculative thinking, creative coding, 3D modeling, and the fundamentals of machine learning and AI. This holistic approach not only expanded my design toolkit but instilled a profound awareness of ethical considerations, shaping me into a designer with a nuanced understanding of responsible and impactful creative practices in the ever-evolving realm of technology.