Close to the Machine
by Anchalee (Joy) Panigabutra-Roberts
I thought about AI (artificial intelligence) and it took me back to a book I read many moons ago by Ellen Ullman, a woman computer programmer, with the title, Close to the Machine: Technophilia and its Discontents (San Francisco: City Lights Books, 1997). It is her memoir as a female computer programmer in a male-dominant profession. She talked about how the coding process consumed her life and how she needed to work long hours and cut off from social life to keep the codes in her head to stay focused on the codes or to be ‘close to the machine’. The book discussed other aspects of a programmer’s life, but this part about the time alone to keep all the codes in her head to stay close to the machine resonates with me. I do work with our library system’s database and I am responsible for over three million MARC records and many workflows to manage. When I need to set up a new project, a process within the system or troubleshooting our database, it usually requires me to work with MARC and other system components to execute the jobs that often time impact large datasets in our system or other repositories such as OCLC WorldCat database, HathiTrust or Center for Research Libraries, for examples.
When I learned how to code in Python, the requirement of quiet concentration and focus time was also at play. Working with codes has many moving parts that are interconnected. If we miss one step or one character in the code, it can throw off the whole sequence and it will fail to execute the codes. This aspect of coding/programming is both a blessing and a curse. And it is how the coding process is designed. I will reflect on this aspect of AI in this blog, as well.
This past week I have had a privilege to be a part of the IDEA Institute on AI and it reminds me of this concept of being close to the machine. Not only that we got to see different uses of AI in the galleries, libraries, museums and archives (GLAMs), we also got to experience learning programming, testing existing AIs and designing an AI or AI-related projects ourselves. This process required us to think ‘like a machine’. That is, we need to understand how a program or an app works to get it to work. We have to understand its structure and its requirements, both in terms of the tools and the data requirements. For an example, vanderbot requires .json and .py files to run the program in the terminal on my computer with the data in .csv format and with the python3 version. The conversation AI in Google requires different components to be filled in with different types of data to get it to work correctly.
One thing I found to be in common when learning about a few library systems I got to work with and also many software applications was that it required me to ‘think’ like the machines. I had to understand the data architecture/designs, the functionalities and the data requirements of each system or app. Understanding all of these components helped me not only to work with the systems and apps effectively, but also for troubleshoot problems that would come along. AI is no exception, since it is built based on sets of instructions and system architecture and design to execute the different functions we design them to do based on our data feeds.
From al-Jazari’s elephant clock to Google’s conversation AI
When I think about al-Jazari and his inventions that dubbed him the father of robotics, I can see the connection between his elephant clock and the conversation AI that we got to have a hands-on design during the institute. As I mentioned above, they are quite amazingly similar. Both have specific design elements and the sequential nature and interconnections of different moving parts of the designs to execute the outcomes. This conceptual design idea behind robotics is going back as far as the 12th Century and is still thriving in different manifestations to the present day. al-Jazari’s elephant clock and the conversation AI have much in common and with a shared history.
My final thoughts on AIs
AIs, not unlike humans, have both capabilities and limitations. We invent them and we are entirely responsible for the outcomes, both positive and negative outcomes. AIs are our ideas and our input into their makings. AIs reflect our thoughts and dreams, our sociocultural, political and worldview assumptions. However we design AIs, our subjectivities and worldviews will be embodied in the AIs we create and will always be a part of who we are.
AIs are us.
Last but not least, I would like to thank the co-PIs, the advisory board, the funders and supporters and all personnel at UTK campus and beyond who helped make this amazing program possible. My deepest gratitude to you all.
I’d also like to thank my fellow colleagues for sharing their work, thoughts and ideas with us during the institute. It is my pleasure to have met you and I will cherish our friendship and would be happy to support your projects in any way I can. All my very best wishes to you.