2023-10-27 Meeting notes
Date
Attendees
Lawrence Eribarne, Chris, Rebecca, Greg, Eske, Melissa, Nick, Michael
Goals
Discussion items
Time | Item | Who | Notes |
---|---|---|---|
Event Planning | All | Summary:
Notes: Logan - Should we delay until 12/1 to allow more time to set up? Chris - Undergraduate students are ready, depends on the objective - maybe need a zoom session to kick off activities. Michael - We have students now, let's direct them - end one today on if they can get together Friday afternoon or next two weeks. Logan - Alternative - Set up a loose management structure to manage it like a coding project. Don't work completely independently, but have some level of peer code review, documentation, project management, and use this as a project management learning opportunity as much as a generative AI development opportunity. Near term goal - Pair student with a shared list. Rebecca - Will need tight management to keep students engaged. Logan - Put time on Friday next week. Send open call for those interested in AI in teaching and learning, going to get together. Logan - Let's grab a time slot and keep an open office during that time to engage students as they come in. Rachel - Should not brand it as a hackathon. Idea of something a week for now works, but rebrand as brainstorming, project matching. Mike - Need to decide what the end result we want is. One piece is to engage students. Do we have projects we would love to see, then share with the students, and then get their feedback, and then develop on focus areas versus things that won't see the light of day. Chris - His attempt at that list is in the wiki link in that chat. Rebecca - Like everything on the list. Share Mike's concern that if specific faculty don't have a plan to use, could sit there. A lot of good uses of AI involves faculty working with TFs. Most support is not coding, but talking through scenarios and feasibility. Should pair faculty with student. What's one way we can pair generative AI into teaching of a course in spring? Chris - Not mutually exclusive, could do both. Let faculty know we are willing to support an AI facilitator to work with you and your teaching team and assign an undergraduate. Separately, if there is a subset that is useful, we can embark on a development project for those. Rebecca - That works, but next week may be fast. Michael - Useful for somebody immediately? So it goes into use. Michael - If we can come up with three class use cases that are needed, would be a strong success case. Logan - Likes the idea of use cases being something that somebody will use. Been resistant to send a broadcast email so far. We could issue one that says we are ramping up, if you have specific things you're trying to do, let us know so we can be participatory, we will match resources to need and see where we are at. Logan - Narrow down, get them in on Friday and pair up with a list of goals. Nichek - If we have more time, need to include grad students or postdocs where this aligns with research. Rahel - Interested in humanities and non-STEM disciplines, hearing anxiety expressed among students looking at disciplines getting eliminated. Would be interesting to see what faculty could imagine and help students understand what AI can or can't do. Miachel - Two different things. Students that are already hacking could be directed to a class. Other disciplines have other needs to understand what it can or cannot do in their worlds, but would be a different kind of project. Gret - Interested could be engaged based on those willing to 'experiment' not necessarily commit to the project. Some we may be able to experiment by the end of the semester. Could be some short term things, j-term classes, and in fall. Create more of an incentive to push on a particular item. Michael - Faculty struggling with change of calendar and launch of spring class. If we aren't already talking to them about it, but may not be enough time. Tamara - On practical side - we have some ideas we want to get students together - Marlon can help with this, he works with students in learning lab. Marlon - We'd be happy, we are set up to hire students and supervise students. Lots of experience with chatbots. Would be excited to get students working on 2 to 3 projects the team thinks is important. Logain - Next Friday - open house idea to encourage interest maybe still makes sense. If we do it, we need a go/no go decision. Michael - We should be targeted. We have a list, students already contacted by Chris, invite them to a meeting we can all make and talk to them, bring friends, use student network that are willing to work, start with that group. Look for matches, then send another call if we need to. It's ok if it doesn’t work as an experiment. Should wait for 6 hours, setup a time everyone can make. Rebecca - Would be cool to have a kickoff. Like the idea, after that organize it all through the learning lab (key decision). Rachel - Format that could work for ideas implemented in time for spring. Rather than piecemeal from kickoff on Friday until spring. Intensive work period during winter session, or invite them a week early for fully focused coding. Marlon - Makes sense and can decide if it is ok for them to work on it remotely. It would also be great to get extra expertise. They can supervise, but need to call on people with coding expertise to supervise the projects. Rachel - Early return could be an incentive for them to participate. Rebecca - Sandbox access, but when working herself, use OpenAI directly. Marlon - Can even use sandbox APIs for hackathon. Logain - Students, faculty only experience is using GPT-3.5 which is misleading. It is transformative for people to easily access more cutting edge models. Logan - Need to confirm who is available on Firday Rebecca - Didn't have large space in SEC, but if 20 people or so, could use SEC. Logan - Let's put a priority on Chris' availability for Friday. Rachel - Need specific information about the activity on what it is to reach out. Logan - Message is coming - talk to us, we have funding and employment available if we have a project we decide to develop. Rachel - A lot of students are engaged in efforts and being paid. Need to take a hard look at the incentive structure. Nick - Part of meeting could be talk about projects, ideate potential new projects, and then some time for folks to do a little hacking just to get started. - maybe a 2 hour long session. Rachel - Student feedback on ideas of a hackathon. Nichel - Could just be initial planning, APIs, etc. Rachel - They will be ahead of us and will be asking for resources. Sandbox access is also a participation. Logan - A helpful organizing tool would be the canvas site and help facilitate access to students. (Action - List of students for sandbox access)
Rebecca - Shouldn’t be a custom rate - would be normal undergrad season rate. Tamara - Learning lab pays $18/hr. Rebecca - It is problematic to pay students differently. Tamara - Will talk to OER, may have funding for this. Logan - As thinking the CA rate - If they are committing time, they may be doing it instead of CA. CA rate $23.50. Could hire students as a CA for the course they are supporting. Marlon - If we can identify 4 or 5 faculty projects to add clarity. Action - Lawrence -- HUIT actions for student sandbox, upload, API, and expose prompt. Logan - Friday afternoon prompt engineering could be good. | |
Action items
- Former user (Deleted) Send Lawrence student names for sandbox access when ready
Copyright © 2024 The President and Fellows of Harvard College * Accessibility * Support * Request Access * Terms of Use