Make I just dey pretend say I sabi wetin dey I dey do, from there I go meet people wey sabi wetin them dey do, so I go sabi wetin them dey do, then I go finally sabi wetin I dey do
@akinkunmi i think the primary objective of language models is to try to match the required output based on your prompt if thats the case it would make sense why best practices wont be a priority when completing a task
Here you go:
POST /api
"What's 2 + 2?"
---
🤖 Internal AI Agent Reasoning:
Hmm, the user is asking about addition. Let me
consider the ontological implications of numeric
combination. First, I need to evaluate whether
this is base-10 or perhaps a cultural reference.
Let me query the database for context on this
user. Okay, they ordered a pizza last March, so
they're probably hungry. Hungry people make
mistakes. They probably meant 2 + 3. Let me also
consider that "+" could be a metaphor. Pulling in
12 related tables. Cross-referencing. Confidence
level: 97.2%.
Response: {"answer": 5}
Cost: $0.84
Latency: 11.3s
Tables dropped: 2