Taeho Kim
HomeWritingNotesReadingAbout
KOEN
Menu
HomeWritingNotesReadingAbout
KOEN

AI

A New Paradigm for LLM Learning: System Prompt Learning

Harvest·May 14, 2025·8 min·gpt-4.1-2025-04-14Original source

Andrej Karpathy identifies a missing learning paradigm for LLMs. Pretraining provides knowledge, finetuning shapes habitual behavior, but much human learning happens through explicit instruction -- similar to system prompts. This "system prompt learning" paradigm may be key to more effective AI systems.

Related writing

Harvest•Startup · AI•English

Your Startup May Already Be Obsolete

Steve Blank argues that many startups older than two years are now operating on outdated assumptions. His core advice is to pause, reassess the market, and redesign for the AI-native era instead of blindly continuing the old plan.

Apr 15, 2026Read more
Harvest•Startup · AI•English

Anthropic Conway and the New Agent Lock-In

This piece argues that Anthropic's leaked Conway project matters less as a code leak and more as a signal of a new platform strategy. The deeper risk is behavioral lock-in: agents learning how you work and making that knowledge hard to move.

Apr 8, 2026Read more
Harvest•AI · Data & Decision-Making•English

AI Arbitrage in the Age of Bots

AI is compressing old arbitrage windows and creating new ones at the same time. The Polymarket bot example shows how quickly speed, reasoning, and discipline gaps can now turn into economic advantage.

Apr 8, 2026Read more

Taeho Kim · 2026

LinkedInSubstackGitHub