Llm As Phd Supervisor

#structure-note

Below I have highlighted the core areas from the ual-description-of-a-supervisor that I need to consider for my project-supervisor-bot:

Written comments or guidance on your work ~={red}(the bot needs to ‘see’ my work) =~
Face-to-Face meetings, the venue and timing of these meetings is a matter of agreement between you and your supervisor(s) ~={red} (dialogue between myself and bot) =~
~={green} Telephone, email or Skype/Facetime or similar communication.

~={red} functionality of the bot that needs to be worked out

**Responsibilities of the Supervisory Team

~={red} Meeting regularly with the student =~at the intervals agreed at the start of the research programme. The normal expectation for full-time students is that at least three formal (i.e. recorded) supervisions must take place each term. During key monitoring stages of the research such as registration and confirmation more regular contact is to be expected.
Ensuring that any discipline/health and safety policies are brought to the attention of the student and explained.
==Discussing any hazards involved in the research work and how these can be dealt with, and ensuring that safety procedures are followed.
~={green}Responding promptly and constructively to written or practical work completed, within the timescale agreed at the start of the project.=~
When a student has an urgent need for advice or guidance, the supervisor should be able to provide it within a reasonable timescale that does not delay the student’s progress.
Drawing the student’s attention to the examination regulations for the presentation and examination of theses (including any practical work) and agreeing a timetable for completion at least a year before submission.

~={yellow}this is information that I need to train the bot on so that is has the required knowledge for the task =~

~={green}core behavior that is entirely different in human to bot interaction =~

Students may reasonably expect all their supervisors to:

Treat them professionally.
Give early advice about the nature of research and the standard expected in order to achieve the degree for which the student is aiming.
Arrange a supervision meeting as soon as possible after enrolment. This would normally be within the first week for full-time students. For part-time students, this may take longer to arrange.
Ensure that the first meeting covers the areas set out in 2. above. If this is not possible, these areas should be covered in another way.
Provide advice on literature sources and other research resources available and provide some directed reading to facilitate discussion and form the basis of a contextual review.
Deal promptly and appropriately with any research problems.
Take an active role in introducing the student to meetings of learned societies, seminars, etc, and to other researchers or experts in the field or relevant artists and designers.
Where the student’s own creative work forms part of the research, and when adequate notice is given, every effort should be made by the supervisor to attend exhibitions, performances etc, or visit studios as appropriate.
Provide advice on the publication and/or exhibition of research work and on the public presentations about the research.
Put the student in touch with specialists inside or outside the University where the student’s work goes outside the supervisor’s expertise. If appropriate, the supervisor should recommend the appointment of such specialists as internal or external supervisors or advisors.
Provide both encouragement and positive advice as well as criticism and negative comment as appropriate.

In line with ethical concerns regarding these megacorporations and privacy issues - using LLM’s this way, particularly within an institution is problematic. This leads me down the rabbit hole of local LLMs and I have this idea of enabling students (and me) to create custom LLMs (project-supervisor-bot + diy-bot-creation)

in the realm of the more absurd/unhinged, I am also thinking about virtual pets as an inspiration, but also more generally how virtual / AI companionship has evolved over time. What is are the use cases that could be important beyond just ‘information exchange’ and traditional ideas of education purposes

How might an LLM be able to mimc this in some way?

How does this line up with other institutional uses of LLM’s?

**How do I get the bot to stay up to date? Is that possible, do I have to keep doing that? - Could it suggest to me updates that are useful, then I can choose to update

It needs memory and context