It’s an understatement to say that artificial intelligence is in peak hype cycle these days, as a hundred million people flock to ChatGPT and billions flood into AI startups. Through newer machine learning models and vast increases in computing power, these new technologies will almost certainly make a large change on the world, especially in how we work. With this knowledge, it would be wise for anyone to at least understand the capabilities of ChatGPT.

For July’s M2M, I will be studying prompt engineering and applying it to an NBA games and odds dataset.

Here’s my mission statement:

Within the next month, I will understand ChatGPT’s capabilities toward generating content about current events outside its training data on which it has no knowledge, such as current NBA games. Signs of my progress will include developing ChatGPT queries which generate the content I am seeking, being able to iterate on queries with NBA games and odds parameters, and applying few shot learning to get the best possible output.

Here’s my action statement:

The activities I will conduct are identifying five variables I can generate and pass to my query, simulating 10+ weekly queries with these parameters, and developing an end product of a newspaper or blog article to analyze a single game.

Project Plan:

• Week 1
  • Write monthly introduction post
  • Identify five variables with game result, box score, and odds data that ChatGPT can use to write an article
  • Write initial queries to see what ChatGPT produces
• Week 2
  • Define prompt’s requirements and constraints to produce desired output format
  • Pass variables to prompt and test output
• Week 3
  • Iterate on work done during Week 1 and Week 2
• Week 4
  • Produce article that takes NBA game data, player information, and odds to produce full game preview
  • Write M2M recap post

I’ll check in at the end of the month. Happy July.