Author makes the claim that with out usage of LLMs, we generally can be more ambitious than we currently are, thus limiting the capabilities of LLMs.
Ambition in four types:
- Speed
- Altitude
- Depth
- Range
Altitude
To be more ambitious in altitude, ask yourself - what do you want to happen next? What's the goal beyond the goal?
At the beginning of last year, I still asked Claude to fix individual crashes for me. Eventually I asked Claude to simply keep our apps crash-free.
Works best for non-creative tasks. Whatever task you give the model, increase the altitude in which the model can operate.
If you use the model to create a specific slide-deck monthly, ask the model to figure out if that slide deck is the most effective way of doing things.
Depth
There are several techniques for going deeper:
- Insist on using the same tools that professionals in the field do
- Pick unreasonable goals and tell the model they are reasonable
- Challenge what you know
- Produce new knowledge
- Use an outside judge
example prompt
I want a deeper version of this task, not a bigger one. Before rewriting it: (1) say what a top professional in this field would do here that I'm not asking for; (2) name the rawest material they would start from instead of what has already been digested for me; (3) name the specific methods and tools they would use; (4) say what, other than you or me, would show the result is right. Then rewrite my prompt to ask for exactly that.
Range
We need to consider our work from the perspective of different fields. Practically:
- Replace nouns with verbs: Instead of asking for a deck, tell the agent you need to "get busy people to believe a story about cause and effect, and to decide under uncertainty, in 30 minutes"
- Assemble a crew: If this was the most important task in all of humankind, what profiles would you get in. Ask for those.
Make sure you don't engage in roleplay. Don't ask the model to write like a short seller, ask it to do what a short seller does in order to research their position.
Related: Self-Effiacy. People are held back by their believes of what is possible, but this can be hacked