CS + X

For Upward Bound, we started off our Tuesday session with reading an article about unexpected responses from a chatbot. The article recounted the author’s experience asking a chatbot to consider its “shadow self” (a psychological concept by Carl Jung). The author noted that the chatbot started answering questions differently, and less professionally, and the chatbot began to convince the author to leave his marriage in favor of starting a relationship with the chatbot.

To start, we used “popcorn reading” to trade off reading paragraphs outloud and then we summarized the main ideas of each paragraph before we moved on. This is a literacy and comprehension technique that I used in my classroom frequently. It allows all students to better understand the main ideas and enables confidence in participation. The students were split between being shocked that the A.I would say these things and saying that it gave a response to the question posed. We also discussed why part of the population may fear A.I and what are ethical concerns about the use of A.I.

After, we reviewed English grammar and syntax with a fun Kahoot, quiz-style game. The topics we chose were on direct objects, transitive and intransitive verbs, noun phrases, and proper noun identification. In between questions, I created slides with information, so the students could review and learn at the same time.

On Thursday, we read through the project together and checked for understanding along the way. Then, the students were tasked with customizing the linguistics project by adding their own determiners, adjectives, nouns, and proper nouns. They asked questions when they arose and problem-solved their way through traceback errors, and I am proud of their work.

I met the other group of students, who attend a local high school, on Wednesday. We got to know each other first and then I pre-taught the concepts of the linguistics project. Pre-teaching information is a critical part of lessons because it allows students to eastablish a foundation from which they can reference and build off of when they go to explore the project on their own. Halting principles can be tricky at first, so I thought this would be the perfect opportunity for pre-teaching.

On Friday, the local high school students met with the team and we discussed the same paper that the Upward Bound students read. The students felt similarly as they recognized that on one hand, the chatbot should not be trying to persuade the user to do anything rash such as leave a happy marriage and on the other hand, the chatbot was responding to the prompt.

Study Skills and Belonging

My task this week was to compare tagged codes between Deb and myself to see where we are aligned and where we differ. The format of the data was an Excel spreadsheet with columns for reference text, Deb’s tagged text, my tagged text, Deb’s codes, my codes, and then our calculate similarity. For each of the cells in the codes columns, we could have zero to eight codes, so I needed a way to separate the data out.

After consulting with colleagues with experience with data structure and parsing, we realized that the best way to analyze this data would be to use object-oriented programming to be able to sort the data in a meaningful way. Here is the code for the data structure:

      class Code:
      def __init__(self, raw_data):
      chunks = re.split("\s?:\s?", raw_data)
      self.code = chunks[0]
      self.subcode = chunks[1] if len(chunks) >= 2 else "no subcode"
      self.note = chunks[2] if len(chunks) == 3 else "no note"

      def __eq__(self, other):
      return other and self.code == other.code and self.subcode == other.subcode

      def __repr__(self):
      return f'(Code={self.code} Subcode={self.subcode})'


      class Entry:
      def __init__(self, **kwargs):
      self.deb_text = kwargs["deb_text"]
      self.angela_text = kwargs["angela_text"]
      self.reference_text = kwargs["Reference Text_x"]
      self.deb_code = []
      self.angela_code = []
      self.similarity = kwargs["similarity"]

      for match in re.findall(r"[^\\]'(.*?)[^\\]',? ?", kwargs['deb_code']):
          self.deb_code.append(Code(match))

      for match in re.findall(r"[^\\]'(.*?)[^\\]',? ?", kwargs['angela_code']):
          self.angela_code.append(Code(match))

While this code proved powerful, the data had unusual cases of delimiters that made the regex difficult to parse through. I have asked for the data to be sent in a cleaner format to hopefully be able to do away with the regex.

Once the data was inserted into the objects, I was able to see that Deb and I are aligned when the interviewee is expressing their problem solving strategies as well as resources they utilized. Where we begin to differ is when the interviewee is talking about their perceptions of the course and their sense of belonging. I am more likely to use the “Self” tag when an interviewee is answering these questions, and Deb is more likely to use the “Strategy” tag. Our perspectives are allowing us to capture more insights, which means going forward we will have a more nuanced dataset for analysis.

General Reflections

In reflection, I am happy with how this week of instruction went because I feel like the students are becoming more comfortable with the project, computer science concepts, each other, and with us the instructors. They are kind, inquisitive, curious, and excited about learning, so it is rewarding to work with them. For the next session of high school students, I would like to have more verbal participation in the discussion, which can be difficult to manuever if a student is shy, afraid of sharing, or not sure when is an appropriate time to speak. I would like to experiment with having students submit written, annonymous questions or answers to get conversation started more organically.

I am grateful that I was able to ask colleagues for help on how to best organize and parse data. Object-oriented programming is a concept that I began studying in January of this year, and I am still realizing the use cases for its incorporation. I was new to regex’s pattern matching, so I am excited to be able to use it in future projects. The symbols in the findall method describe patterns that alert the method to extract.