Preparing talks with the help of large language models

It seems every day there's another news article about how large language models (LLMs) are going to replace hundreds of thousands of jobs.

At least in science, we are safe (for now) because the model outputs cannot be trusted and must be carefully vetted for factual accuracy. Thus, we should be thinking about applying LLMs to tasks in research where accuracy is not essential.

Next week I will be giving a talk on reviewing the scientific literature to the graduate students at CQT. Whenever I have to present a somewhat dry-sounding topic it is nice to begin the talk with a semi-humorous ice-breaker to grab the audience's attention. For example, one might try to relate the topic to a popular movie or book.

This is an area where LLMs can really shine! I quickly obtained some good ideas using the 7B and 13B parameter versions of LLaMA (running locally using llama.cpp). The prompt used is highlighted in red:

• Reviewing scientific literature can be challenging. It's like the movie Groundhog Day, you have to keep going through all the old material before finally getting past it and reading something new. And then having to do it all over again.
• Reviewing the scientific literature can be challenging. It's like the movie Inception; you have to wade through layers of reality before you find what's really going on.
• Reviewing the scientific literature can be challenging. It's like the movie Speed - if you slow down, you die! Rather than read every paper in a specific area of interest, scientists typically search through several papers to find those that relate to their research. But finding the right information requires patience and skill.
  
• Reviewing the scientific literature can be challenging. It's like the movie The Matrix where you have to plug into a computer to find out what is real and what is not.
• Reviewing the physics literature can be challenging. It's like the movie 300, "this is madness." If you read an academic paper or scientific article and you don't understand what it's about, chances are very good that the author didn't either.
• Reviewing the scientific literature can be challenging. It's like the movie 300, where 300 Spartans tried to hold off the Persian army. The odds were bad and the Spartans knew it, but they held off the vastly larger enemy force until they were completely annihilated. That was a glorious defeat and I hope this is one as well. The scientific literature is also vastly larger than your typical researcher can read and understand. One paper I am reviewing has over 600 references, so that's not a very promising start for the scientifically challenged like myself.

For further engagement, the next step can be to take the idea you like the most and use an image generation model (such as AUTOMATIC1111 - now easy to install and run locally!) to make a mash up of the movie and the talk topic. More on this another time!