Sabine Cossette, MIS Individuals & Societies, Theory of Knowledge Teacher and Head of Department welcomed back to campus Thad Hughes, her former Computer Science teaching intern! “Thad was hired to be my computer science class intern in 2003. He was straight out of the University of Virginia. I loved supervising and working with him - his enthusiasm for learning is infectious. Since having been an intern, he has gone back to school, received his master's in computer science at Stanford University, worked for Google for 12 years, and now for a biotechnology startup. He has become extremely knowledgeable in the fields of machine learning, data science, and artificial intelligence”. This was Thad’s third trip back to MIS to do a workshop for Senior School students.
He is currently the Associate Director of Machine Learning and Data Science at AbCellera, a bio-tech research company. Thad provided an interactive presentation using Codelab, focusing on Machine Learning & Data Science for Life Scientists.
Thad presented Theory of Knowledge students with some very profound information affecting their perceptions of ‘data’. “Data sets can also accrue bias”, explains Thad. “Data can only be interpreted with assumptions” which allows for a gateway where human bias leaks into data sets. This in turn affects machine learning. “Machine Learning is the story of causality, while humans are driven by correlation”.
Ms. Cossette confirmed that “Thad raised many important 'ToK moments' regarding data science and how we approach research. One of the big takeaways was to be aware of how bias at the onset of the production of knowledge can constrain the pursuit of knowledge”. As a ToK teacher, Ms. Cossette had no trouble keeping up with the high-level concepts, but did ToK students get the message?
11th grader, Iiris shared that her key takeaway was “how much misunderstanding there is with data, especially that in order to interpret data, you have to have an assumption already. It is generally believed that analyzing data as a human is objective. But the idea that you can’t analyze it without already assuming something, making it less objective, is really interesting”.
Classmate Francesco explained that “data can be interpreted in different ways and the way in which it is interpreted can change what the data represents. It helps you understand that there can be bias in data even though data seems to be the rawest form of knowledge. But it can be biased, so you have to be careful how you use it and how you interpret it”.
11th-grader Finn was thankful for the experience and pointed out that “many MIS students will also have a future in this area, so it’s very good to have this opportunity to learn from Thad”.
Judging from these key takeaways from a sample of students, the presentation was well-received and well-understood. Ms. Cossette explained that the end goal is for TOK students to be able to recognize bias in their own work, including their Internal Assessments and their Extended Essays, and to be able to apply it for life-long learning so that they make epistemically responsible decisions when faced with knowledge on important topics, be it political or otherwise.