1 July Monday - Mentoring Circle

Attendee: Whole group + Rhys

• Rhys advised to create a plan for the project (e.g proposal, review etc), based off of previous LG of PX (available in library). 
•  
• Discussed how good projects could go into Optimise (showcase)
   
• Discussed IP consideration - client has commercial IP, we (team members) have technical expertise/IP so we still have stake, but if we do not want to develop due to this IP consideration, we must let Rhys know immediately and would be taken off this project  and put on another one instead. 
   
  • As technical IP holder, we still are not liable for anything. 
  • Strongly considering using MIT license

By next week:  
Input, output and how to measure success of each user story. 
Including data that we are feeding into the system. 
RAG system development is critical for this - start learning that. Each person should do their own research, and then compare notes with each other. 
** Discussed exciting possibilities - we could possibly create videos as well. 
 

9 July Tuesday - Client Meeting Minutes

Attendee: Whole group + Brahm

• 6 characteristics in a family background

• The characteristics in this slide are often at high risk. 
•  
• Caution with indigenous family: since the risk of hallucination/unpredictability of output, avoid the indigenous family scenarios. We could add a specific guard rail for this. 
  • Oliver posed an idea of creating a detailed context for this? (more work up front, but this could allow us to have a more curated output that would helps us include indigenous families appropriately) 
  •  
• Group to come up with a ballpark figure cost, for Brahm. Brahm is not able to provide an estimate of budget as of yet. 
   
  • We can do around 15 families, so students work in groups (which would help lower the cost of having it for 150 students instead) 
  • Brahm is happy for batch processing, as this is not time critical. 
     
• Briefly discussed alternate versions: 
   
  • Interactive chatbot platform
  • platform creates static pages/output over time, not interactive chatbot
     
• Brahm loves the timeline idea, so students track the progress easily 
   
  • We could get students to work on different families by viewing the timeline (reflecting a real life scenario)
    Students will visit their families 20 times over their degree (4 times a year max).
     
• On the topic of unpredictability/hallucination:
   
  • This can be an opportunity to have students practise better prompting/questioning.
     
  • “How to train your dragon” 
     

15 July Monday - Mentoring Circle

Attendee: Whole group + Rhys

• Rhys suggested use Figma to create a much more presentable prototype, to consolidate and clarify requirements 
•  
• Rhys suggested to show Brahm user stories, so he can interpret them and provide feedback as required (to avoid any misunderstandings) 
• On the topic of students review of AI generation: 
  • Perhaps students could rate the output at the end.
    Dangers of having students rate the AI as we go, is that they could intentionally keep rating it as incorrect when actually it is correct. 
    Gamification opportunity: students could get a bonus point for flagging incorrect/outrageous AI output, which could incentivise students' reporting, which also prompts them to think critically of the patient's history. 
    AI/Invigilator/Student. 
• Cost/Memory:    
  • Rhys agrees with the summaristion idea, as it reflects real world dr/patient interaction for example read from a summary at every consult/workshop
    Export function that converts the history to a pdf (could be submitted to TurnItIn or academic auditing). 
• Must have 80% attendance with Client (min 2 members in a meeting) 
• Week 6 prototype can be the Figma prototype, but really important to get started on coding/make commits to Git repository, to avoid last minute rush. 
• Rhys will set up a GitHub classroom.
• How this project will be assessed: project will be shown to the academic panel + client. Best score (most favourable to students) will be selected. Client still needs to decide how much of the functionality was achieved (under 50% is fail). 
• Team will meet with Rhys for 1 hr each week (outside of club)

Action items for next week: 

• Ask Brahm for data for modelling. Ask him to provide it by the end of the month. If unavailable, ask him how we could create some data to work with.
•  
• Ask for students' past work. Try and get a range of work e.g poor, average, exemplar. So we can understand how the end user would be interpreting the output/ working on it. 

• Add two user stories: Gamification feature and Export PDF function 

   

17 July Wednesday - Group Meeting minutes

Attendee: Whole group 

• Gone over user stories to come up with initial sketches.
  • User story modification: Extra button to indicate the end of the conversation (like ending a consultation) → help in triggering the summary process.
  •  
  • Create Figma project with some initial wireframes. 
•  
• Discuss about prototyping 

Action items for next week:

• Focus on completing wireframes 
•  
• Make progress on project plan and project proposal
• Question to ask/discuss: How and When does AI return the consequences/effects? Should there be another button for students to click to receive feedback immediately or return consequences together with new life events after few weeks?