1. Examples
2. How
2.1. Have right mindset
2.1.1. AI is a helpful assistant
2.1.2. AI is good to elevant info to be reviewed
2.1.3. NOT to make decisions
2.2. Learn (patience)
2.2.1. Clearly define the problem
2.2.2. Give clear instructions
2.2.2.1. Term
2.2.2.2. Type
2.2.2.3. Context
2.2.3. What are the capabilities?
2.2.4. Scalable and Sustainable
2.3. Invest time and Energy
2.4. Start to develop skill
2.4.1. Prompt Engineering
2.4.1.1. Good Communication
2.4.1.2. Avoid Mistakes
2.4.1.2.1. Poor direction
2.4.1.2.2. Missing Examples
2.4.1.2.3. Ignoring Format
2.4.1.2.4. Overlooking evaluations
2.4.1.2.5. Failing to Try Again
2.4.1.3. Framework
2.4.1.3.1. Content
2.4.1.3.2. Guidance
2.4.1.3.3. RTF
2.4.1.4. Meta-Prompting
2.4.1.4.1. Take a prompt and feed it into Chat GPT and ask the tool how best to prompt
2.4.1.4.2. Tools
2.5. Quality Standards
2.6. Training
2.6.1. Build into work
2.6.1.1. Drafting motion
2.6.1.2. Compare drafted responses
2.7. Test and Iterate
3. ChatGPT 4
3.1. Constraint answer based on criteria
3.2. 90% accurate
4. Ability
4.1. Drafting ~Jr Assoc
4.2. Level playing field
4.2.1. Faster
4.2.2. More accruate
4.2.3. more thorough
4.3. Collaboration
5. Considerations
5.1. Privacy & Data Security
5.1.1. Platform needs to keep info secured
5.1.2. Can't be used to train LLM
5.1.3. Choose Professional Grade AI Solutions
5.1.3.1. Have a team with domain expertise
5.1.3.2. Work with partner who's going to be there for support long term
5.1.4. ChatGPT need paid subscription
5.1.5. Data privacy agreement
5.2. Choose Professional Grade
5.2.1. Domain specific
5.2.2. Secure & Private
5.2.3. What is it capable to do?
5.3. Users
5.3.1. Pro Bono
5.3.2. Grant reporter
5.3.3. Attorneys
5.3.4. Pro se / clients / applicants
5.4. Hallucinations
5.5. User
5.5.1. client facing
5.5.2. attorney user
5.5.3. pro bono user
5.6. Accuracy & Fit for Purpose
5.6.1. Accuracy Control
5.6.1.1. Mostly accurate
5.6.1.2. AI is trained on public knowledge
5.6.2. Define Achievable use cases and success metrics
5.6.3. Build a series of test
5.6.3.1. Engage with tech with real use cases
5.6.4. Validate source materials
5.7. Collaboration
5.7.1. Build
5.7.2. Repurpose
5.7.2.1. DACA -> Work Permit