# 🪤 Programming with LLMs - Project

Team ClickB8: `Chris`, `Paul`, `Hannes`

## Meeting Plan

+ [x] `08.05.24` : Meeting in Stabi
+ `14:00-17:00` Gruppenraum 3 : PIN `157153`
+ [ ] `15.05.24` : Meeting in Stabi
+ `15:00-16:00` Gruppenraum 1 : PIN `249529`
+ `16:00-19:00` Gruppenraum 2 : PIN `338827`
+ [ ] `22.05.24` : Meeting in Stabi
+ `15:00-19:00` Gruppenraum 3 : PIN `465826`
+ [ ] `29.05.24` : ?
+ [ ] `05.06.24` : ?

## Guidelines

What we need to answer for our pitch:

1. Short description of idea
> we will build the first intelligent teddy bear
2. What problem / whose problem does that solve?
> .. hm?
4. What makes you think that you might be able to solve / address it?
> What are your starting points? Any prior art you can base this on?
6. How do you know to what extent you have solved it? How do you evaluate the quality of your product?
> the teddy bear can answer all questions.. ?
8. How will you make sure that your system doesn't cause other problems?
> unfortunately, our teddy bear attempts to incite murder every now and then
10. What will you use to build it? What kind of resources (data, compute, time, ...) will you need for that, and do you have them?
> robotics lab, unlimited compute time, child/robot interaction data"...?
12. What do you need to deploy your system?
> ...

## Idea Brainstorming

### General

+ Ideally something we can continue working on after the project seminar
+ Something interesting
+ ...

### Concrete

1. Paper assistant
+ System that helps users to understand scientific literature
+ interactive chat prompts
+ source highlighting
+ relevant reltated papers
+ highlihght phrases in the paper to get a more detailed explanation
+ tailor made prompts defined by the user
> explain it to a five year old
+ `[+]` For sure helpful tool
+ `[-]` Has already implementations that are probably better than what we can achieve by our own
2. Literature recommendation / Music recommendations
+ Find books that fit your interests / are similar to books you liked
+ Useful integration into a book micro social media
> à la letterboxd, goodreads, ...
+ Can be built upon a classic recommendation system
> Like movie / song recommendation with embeddings
+ `[+]` Would have many fun UX/UI opportunities
+ `[-]` Might no solve a major problem appart from finding good books :)
3. Custom "news aggregator"
+ Create summaries and press reviews from selected news pages / blogs / podcasts (in transcribed form)
+ Possibilty to include recommendation system based on similar content and/or personal interests
+ `[-]/[+]` Similar services exist, but it might be interesting to investigate shortcomings/limitations of such a tool
4. Personalized meal planner / health companion
+ Create weekly meal schedule e.g. for fitness, diet, fasting; tailored to your needs
+ Possibly include additional work out plans or physical exercises
+ Integrate with schedule, e.g. Google Cal or iCal
+ Include "reasoning" and explanation why certain food choices were made;
+ `[+]` Actually useful to create such tools in a non-marketing-ish way (which are most health related apps)
+ `[-]` Critical system, i.e. affects physical health; we have to ensure we do not recommend battery acid for breakfast
5. Personalized Week Planner
🐁

## Literature Recommendation Micro Scoial Media

### Guideline Questions

#### 1. Short description of idea
#### 2. What problem / whose problem does that solve?
#### 3. What makes you think that you might be able to solve / address it?
#### 4. How do you know to what extent you have solved it? How do you evaluate the quality of your product?
#### 5. How will you make sure that your system doesn't cause other problems?
#### 6. What will you use to build it? What kind of resources (data, compute, time, ...) will you need for that, and do you have them?
#### 7. What do you need to deploy your system?

### UI sketch


## Todo until 15.05.

- [ ] Read Medium article on recommendation systems: https://archive.ph/k3D0f
- [ ] Answer pitch guideline questions w.r.t. literature recommendation app (see ## Guidelines above)