2023 April 27 Jenkins GSoC Office Hours

Attendees: JMM, Kris Stern, Bruno, Adrien, Yiming, Rajiv, Harsh Pratap Singh, Mukul Kumar, Sonali

Agena & Notes:

  • The mentors have done the grading and rankings as well as forming mentoring teams around the selected projects.
  • We have received over 60 proposals this year, a good proportion of great proposals.
  • Google will decide how many projects among the ones the org admins and mentors have selected to sponsor.
  • Deadline for ranking closes at 18:00 today, and the announcement will be on May 4th.
  • JMM will be reaching out to the candidates whose proposals are not selected but show good potential.
  • Mukul asks: “If a better idea about how to implement the project comes up after the proposals have been submitted, can we still implement the improved and better idea?”
    • JMM: The project should be open to variations, the proposal is not a contract, and is subject to change. Milestones and deliverables are guidelines only, and serve to help facilitate the management of the project.
    • Adrien: The case is similar in industry.
    • JMM: This topic will be discussed further in the Bonding Period.
  • Mukul asks: “What is considered open-source and what is not? Are there clear definitions of this?”
    • JMM: Always do correct attribution. Give credit where it is due and be respectful of other people’s work. Don’t do to others what you do not wish to be done to you. Cannot use plagiarism in this scope.
    • JMM: Unless it is super trivial, if it is a neat trick or a good algorithm then you should at least give attribution via a comment. Exactly the same way for images used in presentations. Do not claim something is yours unless it originated from you as a piece in intellectual property. There are some very specific codes about citation to follow.
    • Kris: It depends on which university you attended whether academic honesty is taught at school or not.
    • JMM: Academic dishonesty is considered cheating at school. The spirit is to encourage originality and fairness.
    • JMM: LLM-based apps like ChatGPT/GPT-4 sometimes take ideas from sources without crediting its source, so it works like a supercharged search engine generating good tips instead of generating new knowledge based on various sources of information found on the internet. If you use info generated by ChatGPT you should still give credits to OpenAI/ChatGPT. Do not make belief or impersonate someone or some entity for the sake of honesty.
    • JMM: Once LLM-based applications can replace human programmers/coders then we will have to think about the implications of a computer as an agent. But at this stage we do not yet have the enough knowledge to know all the ramifications.
    • Adrien: If you start from the code generated by some LLM-based applications and you do not cite, the application can give buggy code that it cannot debug. This liability rests on the user of the code snippets. We are in a sense creating the glue between pieces of code. It is our credibility on the line if we claim some code from ChatGPT is our own, and if this code does not work the way we expect it to. I would have a big problem with it in that case.