The Limits of Imagination : An Open Source journey
summary
A readable inventory of claims tools results and failures: a terms‑and‑conditions pipeline trains on 2,000 hand-annotated sentences to extract obligations permissions and risks and recommends LLM synthetic augmentation to reach ~10,000 examples; a fake‑news system combines RAKE NLTK POS/dependency parsing news APIs and reverse image search and cites 22M interactions and 10M+ tweets from 700k accounts; contradiction detection uses SNLI 570k pairs and GPT2/BERT embeddings; a prototype for mobility uses Arduino + ultrasound sensors + Bluetooth and an Android app — it works as a proof‑of‑concept but needs better UX.
AI-written summary · about
chapters (13)
- 0:00Introduction to the Journey
- 0:18Personal Background and Bio
- 1:17Project 1: Terms and Conditions Summarizer
- 4:20Project 2: Only Bads in the Building
- 11:59Project 3: Fake News Detector
- 16:20Project 4: If Your Friends Are Bullshitting
- 27:24Project 5: Sentiment Analysis in News Headlines
- 28:05Project 6: Gender Bias in Educational Materials
- 31:14Project 7: Chatting with ChatGPT
- 36:02Project 8: Experimenting with SVMs
- 39:38Project 9: Shoes for the Visually Impaired
- 40:50Project 10: Creating Your Own Data
- 43:19Conclusion and Final Thoughts
related talks
45mVenturing out into the world with AI
01h 03mIt's All About AI
42mApplying Responsible AI with Open-Source Tools
40mBeyond the Hype: Realistic Expectations of AI
34mTerrible tokenizer troubles in large language models
39mNarrated Reality
17mNatural Language Processing in Automation
55mDemystifying Natural Language Processing across several industry verticals