I mentioned last time that the world seems to have trichotomized:
- Those who have not tried ChatGPT.
- Those who in haste labeled it a chatbot.
- There is also a third class, one that I belong to.
For me, ChatGPT has become my closest collaborator. Perhaps I might merely be a forerunner on humanity’s march towards the inevitable technological singularity? You know, that special moment when we humans finally discard our keyboards and mindmeld with our favourite chatbot.
Anyway, if you’ve ever taught large university classes, you’re likely familiar with reference letter requests from past students. Personally, no matter how busy I might be, I always feel compelled to respond if a student has excelled in my course series or has completed a summer project under my supervision. I of course also routinely receive similar requests from more senior colleagues and a diverse range of others. Indeed, for better or worse, the entire academic system appears to be greased by reference letters.
Accordion Sample Effective Use of ChatGPT: Prompt Engineering for BibTeX Citations by Self-Prompt GenerationTitle
The world seems to have split into three parts: those who have not tried ChatGPT, those who asked it a vague or extremely difficult question and labeled it a chatbot, and a third class, which I fall into. For me, ChatGPT has become my closest collaborator.
We are entering a new era of humanity where artificial intelligence is used to help many of us improve ourselves. This tool has helped me overcome some of my own inherent weaknesses, such as not taking the 6,000 years without sleep a human would need to read the 800 billion words of text ChatGPT was trained on. Obtaining outstanding results is not entirely trivial. With some effort, I have used the tool to improve everything from Mathematica programs to grant applications.
I use ChatGPT-4. This costs me 20.00 USD a month. However, you would be able to do what I am saying here with the free version. You can also experiment with Google Bard, which is new and not as smart yet. I should mention that ChatGPT-4 powers the new Bing, where you can chat with it for free. But I have not done that yet. I simply log in to the OpenAI interface and start a ChatGPT-4 cycle instance. I typically enable the browsing plugin, which is only accessible if you pay the $20.00 subscription fee.
The ability to predict future events lies at the heart of physics. Ever since its inception, a key goal of the discipline has been to forecast the future state of a physical system based on current knowledge. In my work on stochastic and quantum systems, it’s clear that a system’s current state doesn’t always deterministically map to a future state.
Rather than diving into equations of motion today, we’re going to explore a handy risk analysis tool. Regrettably, I’m unaware of the official term for this method, or where it’s most commonly applied. So, if you’re familiar with this approach, please enlighten us in the comments. This method bears some resemblance to the doomsday argument and employs the Copernican principle, both of which we’ll touch on later in this post. I recently brought this line of thinking up when visiting Tim Byrnes at the NYU-ECNU Institute of Physics: the discussion was fun enough that I figured its worth presenting it here.