GPT-3 wrote the description for this talk... Scary or exciting?
summary
Shows how to wire Python to OpenAI and AI21 APIs and DALL·E, using zero-shot and few-shot prompts, prompt cleaning, optional fine-tuning and a FastAPI backend; temperature tuning controls creativity versus predictability. Describes building a "docs-to-talks" web app (about 20 hours front-end work) that generates plausible conference proposals and image outputs, compares OpenAI vs AI21 results, and flags real risks: hallucinations, large-scale misinformation and targeted manipulation.
AI-written summary · about
chapters (11)
- 0:00Introduction to AI language models
- 1:05Personal story and GPT-3 proposal
- 3:14Human impact of language models
- 5:25Other AI models beyond GPT-3
- 7:54Building a talk proposal generator
- 11:13Technical implementation details
- 13:43Sponsors and model parameters
- 14:26Demo of the unicorn app
- 16:34Risks and ethical concerns
- 18:23Potential benefits of AI tools
- 19:06Q&A on evaluation and intent
related talks
51mOpenAI and the AI that may overtake humans
59mKeynote: The past, present, and future of AI for application developers
08mInterview with Lars Rinnan
55mAI Showdown: The Good, The Glitchy, and The Risky
45mBringing the Power of AI to Your Application
32mHow to Make AI Work for You: A Deep Dive Guide into ChatGPT for IT Professionals
58mKeynote: The past, present, and future of AI for application developers
32m"How Smart is AI? – Real projects to build real understanding"