Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

Video course · 11 chapters · 70 min · certificate

Self-Improving AI Agents: From Scaling Laws to Agentic Systems

From scaling laws to agentic systems: how LLMs improve with scale, emergent reasoning, fine-tuning and alignment, test-time scaling, self-improvement with verifiers, and research agents that act as AI scientists.

Become a memberAlready a member? Sign in

What you'll learn

  • Explain scaling laws in LLMs
  • Describe emergent reasoning and chain-of-thought
  • Outline the fine-tuning and alignment pipeline
  • Explain test-time scaling and repeated sampling
  • Connect verifiers to self-improvement
  • Describe research agents and AI scientists

Chapters

11 chapters · 69:42
  1. 2:00 01Intro Members Course introduction Welcome to a course on self-improving AI agents.
  2. 5:48 02Scaling Members Scaling laws Models improve predictably as parameters, data and compute grow.
  3. 5:18 03Reasoning Members Emergent reasoning Bigger models show emergent abilities like chain-of-thought reasoning.
  4. 6:54 04Pipeline Members Fine-tuning and alignment Pre-training, instruction tuning and alignment turn a base model into an assistant.
  5. 10:00 05Test-time Members Test-time scaling and monkeys Repeated sampling: enough monkeys eventually type Shakespeare, and enough samples solve hard problems.
  6. 10:47 06Self-improvement Members Scaling at inference and self-improvement Bringing test-time gains back into training is the heart of self-improvement.
  7. 9:13 07Agents Members From chatbots to agents Chatbots and reasoning models are still single-turn; agents act over many steps.
  8. 7:03 08Verifiers Members Self-improvement and verifiers Verifiers like generated unit tests provide the feedback loop for improvement.
  9. 4:26 09Research Members Research agents and AI scientists Agents that generate ideas and iterate on experiments as AI scientists.
  10. 1:05 10Logistics Members Logistics and assignments Prerequisites, assignments and projects.
  11. 7:08 11Project Members Final project The course project: build agentic systems or go deep on a research question.