Skip to content

LLM Assistant

⬅️ [Neural Manifolds](<./Neural Manifolds.md>) | ⬆️ [Ideas](<./README.md>) | [Learn to create ambiguity using an adversarial cycle system](<./Learn to create ambiguity using an adversarial cycle system.md>) ➡️

Gemini API Key
AIzaSyCRIK_N9MnUwWsCgv4JNrrlr-BJsMvjS2I

Desired capabilities

Keep me updated on talks at my university

I am interested in talks that are stuff that I do, but in very different fields. So anything with interesting machine learning or robotics getting applied to things like chemistry and biology and geology.
Lots of departments have RSS feeds that can give the talks directly. Some have event websites that can be scraped for the talks.
Flow:
1. I feed in a rss feed or website
2. We pull that feed/website
3. A large AI writes a parser that either parses the RSS directly or uses BS4 to scrape the relevant info from the website
4. Every day we scrape each and check for any new talks
1. If the parser fails or we get an empty response, alert so that we can check if the format changed
5. Pass new talks to a large model to assess interest based on a list of example talks with corresponding interest ratings that I pre-curate.
6. If talks exceed a level of interest, send a GC invite directly to me.

BME Events
https://events.umich.edu/group/bme

Robotics Events
https://robotics.umich.edu/events/robotics-pathways-speaker-series/
https://robotics.umich.edu/events/robotics-seminar-series/
https://mcommunity.umich.edu/group/Robotics%20Events

Geology Events
https://events.umich.edu/group/3170

HCI
https://mcommunity.umich.edu/group/Human-Computer%20Interaction%20Seminar

https://mcommunity.umich.edu/group/Quantitative%20Biology%20Seminar
https://mcommunity.umich.edu/group/EARTH%20Geobiology%20Seminar

Periodically look for updates on things:

  1. I put in a name -> it looks up relevant sites -> it proposes which sites to keep an eye on -> it keeps an eye on those sites at a given frequency and also periodically looks up relevant sites again.
  2. I put in a link -> it keeps an eye on just that site

So lets say I like an artist. They have a personal site and a twitter. I enter both of those links. Every N days it scrapes the sites and checks for new content.


⬅️ [Neural Manifolds](<./Neural Manifolds.md>) | ⬆️ [Ideas](<./README.md>) | [Learn to create ambiguity using an adversarial cycle system](<./Learn to create ambiguity using an adversarial cycle system.md>) ➡️