- AI that can create novel content and ideas
- Powered by machine learning models
- Word embedding space-> connect similar words -> word2Vec -> Natural Language Processing (NLP)
- text summarization
- sentiment analysis
- speech language understanding
- Where is the data coming from?
- Is it biased/does it reinforce bias?
- Where do humans fit in?
There are algorithmic bias even in "unfeeling" algorithms
AI wolves - human centered or efficiency centered? We have to define good decisions! -> https://onezero.medium.com/the-ai-wolf-that-preferred-suicide-over-eating-sheep-49edced3c710
- Does the problem merit an AI solution? What is the cost?
- Do you have enough data and is the model capturing desired variables?
- Is the outcome accurate, explainable, and interpretable
| Title | Author |
|---|---|
| Attention is All You Need (transformers for NLP) | Vaswani, Shazeer, Parmar, Uszkoreit, Jones, Gomez, Kaiser, Polosukhin |