This page contains notes, further readings, and general references for the associated lecture.
The Lecture Slides for the talk are also available (use your arrow keys to advance the slides) or download a pdf of the slides.
If you’re looking for notes to the other lectures, head here.
AI Energy and Water Use
Lot’s of different numbers were quoted in lecture. Here are some of the sources I pulled those numbers from:
- Google Gemini energy usage from Google, so good for transparency but you have to trust the source
- MIT Tech Review article on AI Energy Use
- General energy consumption information from the US Energy Information Administration (EIA) on AC and lots of other stuff
- An excellent report on the energy consumption for Data Centers with the breakdown of energy compute vs. HVAC and the different cooling techniques used.
- Goldman Sachs article on the estimated spending on AI for 2026
- Wikipedia page on Jevons Paradox is quite good for getting a basic understanding of it
- Water consumption estimates for various things from the USGS and hamburger consumption estimates for the US
- Good article summarizing the state of AI and water consumption
AI and Copyright
The copyright fight is a complex issue and I am not a lawyer, but here’s a few references for you to read more about the various lawsuits and some of their complexities:
- The New York Times sues OpenAI for copyright violation and OpenAI’s response
- Anthropic settled their suit for $1.5 billion
- Google is also being sued for copyright
- Wired has a good article showing most of the current AI lawsuits
AI Alignment
The alignment problem is a big deal. It’s fundamental and existential. Several prominent AI researchers (including Nobel laureate Geoffery Hinton) are deeply concerned about AI alignment. It’s a wide open problem and people are actively working on it, though if I had to summarize current progress it would be “we will figure it out live”, which is not a great vibe.
- Nick Bostrom writes quite a bit about it
- Richard Ngo goes into detail about misaligned AIs, originally inspired by Elezer Yudkowsky (who recently wrote a new book about it)
Some AI Usage Tips
- Google has a free Prompt Engineering Guide that is pretty complete
- Google’s AI Studio is a free playground for experimenting with AI models. It’s probably the best place to play with all the different developer level AI tools without getting into actual code. To be clear, all models allow these fine grained controls, but you usually only can access them by writing computer code.
- While we didn’t have time to talk about Agentic AI much, it’s going to be a dominant strategy for using AI in the future (and right now) even though they come with some big risks
- There are several music generators out there, but we looked at Suno during lecture, which does have a limited free tier usage
- Matt Shumer wrote a recent article stressing the importance of learning how to use AI now and I agree