Understanding AI: from Philosophical Underpinnings to Practical Tips

Course Description

From fire to reading and writing to social media, technology changes us when adopted and used widely. Large language models (LLMs) and AI tools more broadly are no exception. So what are the philosophical assumptions embedded in the DNA of modern AI tehcnology? How is this technological turn similar to or different from the past ones? How is AI influencing us today and how might it influence us in the future? Which of these influences do we want to embrace, which ones do we want to counteract? And, most importantly, how to use AI wisely ans skillfully based on waht we know? This is what this class is about.

Format

We will discuss these issues through the class, each of our meetings focusing on one or two more specific questions like

• Do LLMs help you learn and acquire skills? Or do they make you dumber and less creative?
• How often do they make up stuff and what can you do about it? What are their limitations and strengths?
• Should you worry about conversations with LLMs making you psychotic? Or are they a great substitute for a therapist?
• How are AI tools used in scams and how to suss them out?
• What are socioeconomic implications and costs of the increased AI use?
• Is an AI god going to turn us all into paperclips in two years? Or is the hype bubble about to burst and we should focus more on other dangers of AI?

Each meeting will take around two hours from 12pm to 2pm with a break around midway through.

Expectations

No technical prerequisites required, only curiosity! You don't need extensive technical knowledge about AI to take the class: even if you've ever only heard about LLMs and other AI tools, but and are curious about the questions above, you're ready to participate. We will cover all the material necesary to understand the topic in class.

Doing about 30 min-1 hour of reading a week. In order to enrich our intuitions about some topics we talk about, I will occasionally ask you to read something before our meetings. Don't worry! The assigned readings won't ever be technical -- typically they will be famous sci-fi stories -- and won't require more than 30 minutes to an hour to complete. It's OK not to finish a reading from time to time, life happens, but I expect that you won't make a habit of it.

The field is changing fast and there is a lot we don't yet know. Much remains unknown about LLMs and AI technologies since their widespread adoption is very recent. We will keep these uncertainties in focus throughout the class. However, even with this in mind, we will discuss and test existing strategies to use LLMs and AI intelligently.

Your Instructor

Arthur has a background in social science research and is deeply curious about how we can think in principle, how we think in practice, and how to extend the ways in which we can think. He has used LLMs extensively in data science work and research and has been following AI years before ChatGPT was released.