Three serious players. All genuinely capable. Which one fits you is partly workflow, partly taste — but use a paid version or don't bother.
The first wave was conversational — ask, receive, copy-paste. That era is ending. The more consequential shift: AI that does work, not just talks about it.
- You prompt, it responds
- You copy output manually
- One exchange at a time
- Passive — you do the work
- Reads your files, writes new ones
- Runs code, checks its own work
- Works in the background while you don't
- Chains steps toward a goal autonomously
Two tools that make this concrete — both available now:
Think of these current AI tools as an extremely smart, very capable assistant who is also extremely literal. These tools are not search engines — they are collaborators who need very clear direction. The more you use them, the easier it gets to get them to do what you want.
Running guides and tools for PCL faculty — updated as the landscape evolves.
-
AI at Penn Law Guides, tool recs, and policy updates pennlaw.link/ai-resources→
-
Faculty Pedagogy Resources Teaching-specific AI guides and tools pennlaw.link/pedagogy→
-
Custom Faculty AI Skills Open-source tools I built — exams, class prep, memos, email, review. Work with Claude Code and ChatGPT.→
-
Claude 101 — Free Introductory Course Anthropic's official overview — prompting, capabilities, use cases anthropic.skilljar.com/claude-101→
-
OpenAI Academy — Free Courses OpenAI's learning platform — ChatGPT basics, prompting, use cases academy.openai.com→
-
Harvey at Penn Carey Law Log in with your Penn Law credentials app.harvey.ai→
-
Mollick: “Which AI to Use in the Agentic Era” Recommended reading — models, apps, and harnesses oneusefulthing.org→