By Nikki Rademaker, and Yanna Smid
This project was created for the AI in Society course. The final assignment was to reimagine how AI could be different and better in a specific domain, like healthcare or the environment. We had to critique current AI problems and create a playful or hopeful proposal that challenges the status quo. The assignment included making a proposal, giving a short presentation, and writing an individual report explaining the idea and its impact. I chose healthcare and medicine. My goal was to critique the current problems within the healthcare domain, and also with AI. Together with Yanna, I wanted to reimagine a hopeful, alternative vision.
AI is used more often in healthcare. It speeds up diagnoses and helps find new medicines. However, it is often trained on biased data. Many medical datasets focus on white males. This leads to healthcare systems that don't work for everyone. Another issue is that drug discovery still relies on animal testing, even though it is often inaccurate and unethical. Companies also keep data private, limiting collaboration and transparency.
We reimagined an AI system that replaces animal testing. Instead of using animals, the AI simulates how drugs would work inside diverse human bodies. We envision it would use biological models based on natural computing and brain-inspired neural networks. Importantly, the training data includes people from all groups, considering sex, race, and age.
The system also follows FAIR data principles. This means research data should be findable, accessible, interoperable, and reusable. It encourages collaboration between researchers, instead of keeping the data private for profit. This makes drug development more ethical, fair, and trustworthy.
To show our idea, we created a short interactive video/animation. This was created in Episode Interactive that has its own scripting language. AI was used for the voice overs. The video follows a researcher who builds this envisioned AI system. Together with the other researchers, she decides whether to accept or reject the new AI system for testing medicine.
In reality, the watcher gets to decide the outcome of this video. If they accept the model, they should let the video play. If they reject, they should skip the video ahead to 4:45 to see the different ending.
For more information about the project, you can download my full report here.