Critical Response #2 Power to the People

Aaron H.
Music 356A , Stanford University

I’m quite excited knowing that there is work being done to reevaluate how we consider integrating AI and ML into our daily lives. I think when I previously responded to the HAI article, I was most concerned about the people involved in making the decision for the algorithms. There always appears to be some expert group or “skilled practioning” group that great influence the outcome of AI products. They control the input data, what the labels are, as well as reiterate time in again what the expected output should be. As a result, the outcome is directly impacted by what is and is not included in the model. To me, this leaves out one of the most important contributors to technology, which are the consumers. These “end users” are the ones who AI is meant to help, so even if with the help of experts, I always wondered why there wasn’t as much effort going into recruiting the end user to help improve the algorithm. Don’t get me wrong, I understood the point of having expert listeners and practitioners of AI who know the limitations of the technology and have valuable experience. However, the needs of each end user may be different which call for more interactive AI. I believe this will also lead to more transparency and less anxiety about how this technology will affect our daily lives. I also think it would encourage us to be more creative and augment our productivity. It was reassuring to see how something can be made less of a “black box” and how the end user can control what’s most important to them. For example, I found this particularly useful when I was able to change the weights of the spectral features when creating my featured artist mosaic. It allowed me to give more focus on the content I wanted out of the model such as percussive sound. I found the preferences to certain errors also very interesting. I could imagine experimenting with this and tailoring the output sound to sway to a certain direction. With all these different applications, I think developing a common language will be even more important. That way there can be more interdisciplinary work and easier way to communicate with people in different disciplines. Overall, I think I’m quite excited especially with the lens of interactive AI. It makes me feel less pessimistic about being left out in the AI and ML world. I feel like there is opportunity to get involved regardless of what field I end up in.

Tasks to benefit from interactive AI:

Language learning- being able to learn at your own pace and have a conversation based on topic of choice

Accompaniment- Be able to have an instrumental or vocal accompaniment when playing a piece or trying to improv

Course Schedule Selection- AI gives you advice on how potential class can help you advance in your degree based on classes already taken

Gym buddy – Takes notes of form during exercise and “spots” for you at the gym

Internet Search- Grants the ability to selectively choose the bias of search results not based primarily on area

Paint Along- Tells you to correct brush strokes to create the artwork of your choice

Navigate- AI acts as your eyes and ears and can lead you around a city

What’s that tune- be able to hum into device to remember that song that’s been stuck in you hear

Video Game companion- Play online with a bot with different difficulty levels and rank

Nature Guide: Scan plants and wildlife around you and have a nature travel guide by your side