🤖

AI Overview 2025

Jun 21, 2025

Overview

This lecture provides a comprehensive overview of artificial intelligence (AI) in 2025, covering key concepts, prompting techniques, AI agents, AI-assisted coding ("vibe coding"), and emerging trends.

AI Basics & Key Terms

  • Artificial intelligence (AI) refers to computer programs that perform cognitive tasks associated with human intelligence.
  • Traditional AI includes algorithms like Google search and YouTube recommendations.
  • Generative AI is a subset of AI that can generate new content such as text, images, audio, and video.
  • Large Language Models (LLMs) process and produce text; examples include OpenAI's GPT, Google's Gemini, and Anthropic's Claude.
  • Many AI models today are multimodal, able to handle text, images, audio, and video inputs/outputs.

Prompting Techniques

  • Prompting means giving AI specific instructions to achieve desired outcomes, using text, images, audio, video, or code.
  • Effective prompting is essential for advanced AI use; it's foundational for all other AI skills.
  • The "Tiny Crabs Ride Enormous Iguanas" framework: Task, Context, Resources, Evaluate, Iterate.
  • Add persona, desired format, specific context, and example references to improve prompt results.
  • The "Ramen Saves Tragic Idiots" framework: revisit the previous framework, separate prompts into short sentences, try different phrasings, and add constraints for specificity.
  • Use online prompt generators for various LLMs to speed up prompt crafting.

AI Agents

  • AI agents are autonomous software systems using AI to achieve user goals and complete tasks.
  • Examples: customer service agents, coding agents that build applications.
  • Six core components of an AI agent (per OpenAI): Model (intelligence), Tools (actions), Knowledge & Memory (context and recall), Audio/Speech (natural interaction), Guardrails (safety), Orchestration (deployment and monitoring).
  • Prompts are crucial for agent consistency, especially in multi-agent systems.
  • Key development tools: nend, gumloop, OpenAI's agents SDK, Google's ADK, and Claude Code SDK.
  • Multi-agent systems provide specialization; MCP protocol enables universal tool/knowledge access for agents.

AI-Assisted Coding ("Vibe Coding")

  • Vibe coding: using AI to build applications by describing desired outcomes rather than writing code from scratch.
  • Five-step framework "Tiny Ferrets Carry Dangerous Code": Thinking, Frameworks, Checkpoints (version control), Debugging, Context.
  • Always define project goals clearly (PRD), use appropriate tools/frameworks, implement version control, debug methodically, and provide detailed context.
  • Various tools suit different experience levels: Lovable, VZero, Bolt, Replit, Firebase Studio, Windsurf, Cursor, Cloud Code.

Emerging Trends in AI (2025 and Beyond)

  • Key trend: integrating AI into existing workflows and products for productivity and reduced costs.
  • Lower entry barriers for non-coders; productivity boosts for developers, especially through command-line tools.
  • Major growth in AI agents for personalizing and automating services.
  • Staying current with fundamental AI concepts is more sustainable than chasing weekly tool/model updates.

Key Terms & Definitions

  • Artificial Intelligence (AI) — Programs capable of performing cognitive, human-like tasks.
  • Generative AI — AI generating original content (text, images, audio, video).
  • Large Language Model (LLM) — AI models processing and producing human-like text.
  • Prompting — Crafting instructions for AI to achieve desired responses or results.
  • AI Agent — Autonomous software using AI to fulfill user goals with minimal input.
  • MCP Protocol — Standardized agent access to tools and knowledge.
  • Vibe Coding — Building applications through conversational instructions to AI rather than manual coding.
  • Version Control — Systems (like Git) to track and manage code changes.

Action Items / Next Steps

  • Complete the embedded assessments to test your understanding.
  • Explore recommended resources and videos for deeper dives into prompting, agents, and coding tools.
  • Practice crafting prompts using the frameworks discussed.
  • Experiment with beginner and intermediate AI coding tools to apply vibe coding principles.
  • Learn basic version control if working on code-based projects.

Full transcript