Using Generative AI
Download as PPTX, PDF0 likes1,220 views
The document discusses using generative AI to improve learning products by making them better, stronger, and faster. It provides examples of using generative models for game creation, runtime design, and postmortem data analysis. It also addresses ethics and copyright challenges and considers generative AI as both a tool and potential friend. The document explores what models are, how they work, examples of applications, and resources for staying up to date on generative AI advances.
Using Generative AI
- 1. Using Generative AI Making learning products better, stronger, faster
- 2. Mark DeLoura Level Up Games Neil Heffernan ASSISTments / WPI Steve Ritter Carnegie Learning Julia Winter Alchemie Solutions
- 3. Using GenAI for games – For game creation: tools – For runtime: design – For postmortem: data analysis
- 4. What about ethics and copyright challenges? – Some newer models scrub copyrighted material from training data – Some providers are indemnifying users – Open ethics questions about effect on employees
- 5. AI is tool, AI is friend
- 6. A new Clippy? (Bing Image Creator - Prompt Perfect version)
- 7. How does it work? – Text → magic → text reply – Text → magic → image – Text → magic → ?? – Image → magic → image – ?? → magic → ??
- 9. What is a model? Source: On the Opportunities and Risks of Foundation Models
- 10. What is a model? – A large interconnected neural network with weights and biases – Created by training the net on a massive dataset – GPT-3 has 175 billion parameters – Llama 2 has 7B, 13B, 70B models
- 11. Model size – GPT-3 is 175 billion parameters – Trained on 45TB of text data – Parameters are float32: 4 bytes – 175B x 4 bytes = 700GB – Llama 2 model 7B = 28GB – My Nvidia 3070 has 6GB – Making models accessible – Cloud compute (GPU, TPU) – Pruning – Quantization (GPT-3 uses float16)
- 12. Using models – Accessible: Chat – Programmatic: API – Custom: Fine tuning
- 14. GenAI is a tool – Great at summarizing – Great at mimicking – Great at mixing – Great at generating language – (whatever that is)
- 15. GenAI is a friend but… questionably trustworthy
- 17. Example: Generating new assets from existing assets Scenario.com
- 18. Example: Rewriting story problems OpenAI ChatGPT-4
- 21. How do I keep up? – Hugging Face as community gathering place – “Two Minute Papers” YouTube channel for AI research trends – FutureTools.io for latest tools and news – Matt Wolfe weekly AI recap: https://www.youtube.com/@mreflow
- 22. Mark DeLoura Level Up Games Neil Heffernan ASSISTments / WPI Steve Ritter Carnegie Learning Julia Winter Alchemie Solutions