Yuan-Sen Ting (OSU)
How Do I Cut My Project Work Time in Half Every Semester?
This talk focuses on large language models
Artificial Intelligence
Machine Learning
Deep Learning / Neural Networks
Large language models / a.k.a. "chatbots"
Why large language models are a completely new paradigm for scientific research
Humans have already used automation for a long time, but we were still the 'connectors'
Human already uses "automation" for a long time, but human was still the "connectors"
LLMs serve as the glue connecting many disparate parts—historically, a 'human' job
Pursuitworthiness: Rage
Anatomy of a 4-Day Project
Pre-LLM Era: 1 Project ≈ 1 Year
Ideation
~ few months
Data collection
~ few months
Coding / calculation
~ few months
Writing
~ few months
Creative writing
~ 2 days
- Correcting grammar
- Improving structures
- Formatting
- References
(2021 - now)
Writing
Problem:
Does the 'creative' part itself define the author?
What do we even mean?
Breaking down the questions
~ 2 days
- Writing code
- Code optimization
- Debugging
- Cloud integration / deployment
(2024 - now)
Coding / Data
- Data curation
Human-AI interface
Your local
directory
Codes
+ web search
Human already uses "automation" for a long time, but human was still the "connectors"
LLMs serve as the glue connecting many disparate parts—historically, a 'human' job
It feels like
My Recent Lecture Materials - NASA AI/ML Interest Group
https://tingyuansen.github.io/
NASA_AI_ML_STIG/
Ideation
~ few months
"Creative"
writing
~ 1 day
Refinement, ~ 0.5 days with AI
Vibe coding
~ 2 days
Over 10,000 lines of code;
None written manually
Data collection (completed before)
Data curation
~ 0.5 days with AI coding
Ideation
"Creative"
writing
~ 1 day
Refinement, ~ 0.5 days with AI
Vibe coding
~ 2 days
Over 10,000 lines of code;
None written manually
Data collection (completed before)
Data curation
~ 0.5 days with AI coding
Anatomy of the 4-Day Project
If You Think About It... What Did I Actually Do?
Ideation doesn't have a fixed timeline—it happens in the background
"Creative"
writing
Vibe coding
Data collection (completed before)
Practical problem:
What's 'left' is also hard to quantify in the current academic reward system
"Creative writing"
"Problem solving"
"Ideation"
The Seeding Problem:
AI amplifies expertise, writing or coding — but you need seeds first
But is That Even True? It's Never Either-Or
Creative writing
Breaking down the questions
The 'seed' threshold keeps dropping
But what
makes you,
you?
YST, Nature Astronomy, 2025
YST, Nature Astronomy, 2025
Pursuitworthiness
Hypothesis
/ Generation
Pursuitworthiness
"Acceptance" / Experiment
Philosophy Conference
By Yuan-Sen Ting
