AI Papers Review - LLM prompting structure and design - The Art and Science of Speaking to AI - S03
Interested in AI Agents? Join PolyAgent Community:
https://www.polyagent.co/
Overview of the Papers Discussed
📑 Paper 1: "LangGPT: Rethinking Structured Reusable Prompt Design Framework for LLMs from the Programming Language"
https://arxiv.org/pdf/2402.16929.pdf
🔄 Bonus Reading: "Emotional Manipulation Through Prompt Engineering Amplifies Disinformation Generation in AI Large Language Models"
https://arxiv.org/pdf/2403.03550.pdf
Timestamps:
0:00 - Quick introductions of participants
5:35 - Overview of LangGPT paper and goals for the session
10:16 - Participant shares experience trying to understand paper
11:03 - Using ChatGPT to better understand complex topics
16:33 - Live demonstration of using LangGPT framework for spirituality prompts
22:37 - Discussion on challenges of communicating desired outputs to AI
27:03 - Ideas for using AI to enhance communication and update others' theory of mind
29:11 - Potential of AI to help with mental health and behavioral issues
31:39 - Participant shares experience with prompt engineering for spam classification
34:04 - Important considerations when providing examples in prompts
37:25 - Question on using prompt engineering for biological data analysis
43:34 - Response on current AI capabilities and areas for development in analyzing research papers
45:35 - Project to fine-tune language model for Ukrainian and observed phenomenon
48:24 - Participant shares experience and research on bilingual language representation in the brain
50:19 - Additional examples and discussion on mixing languages and language evolution
52:31 - Request for details on research agent and community involvement
53:27 - Participant shares anecdote on mixing languages during German test
54:37 - Suggestions for incorporating text chat during sessions
56:10 - Participant shares experience improving cancer data models with techniques like encoding
57:10 - Wrap-up and closing remarks