1.1 Why AI, Why Now¶
You've heard "AI is changing everything" so many times it's become wallpaper. This lesson strips the hype and shows you the three concrete things that changed — because when you can name what changed, you can position yourself on the right side of it.
What you will learn
- Name three shifts that distinguish current agent tools from many earlier AI products.
- Explain why some software-building tasks became cheaper and faster — and what still costs real judgment.
- Recognize the difference between AI hype and AI leverage.
Builder principle
Start with a bounded experiment. Evidence from your own task is more useful than being early to a trend.
What actually changed¶
AI has gone through repeated periods of high expectation. Three observable changes make current agent tools worth testing:
1. The models got broadly useful. Much older deployed AI was usually built around a narrow task — recognize a face, recommend a song. Large language models can attempt many language-mediated tasks: writing, summarizing, planning, coding, analysis, and translation. Breadth is not universal competence; performance still varies by task, language, context, and stakes.
2. The models got tool access. Current agents can read and write files, run commands, call tools, and inspect results when a harness grants those capabilities. The exact products and timelines change; the durable distinction is between generating a response and taking a tool-mediated action.
3. The cost curve changed. Drafting, scaffolding, and iteration can take much less time with an agent. Compute may be inexpensive compared with professional labor, but discovery, review, security, maintenance, infrastructure, and accountability still cost real money and skill. The practical question becomes: which parts can I accelerate without hiding the risk or shifting the cost downstream?
Why this matters to you specifically¶
Language is a powerful new on-ramp. Clear description and honest verification let more people participate, while domain knowledge, accessibility, time, money, tooling, and technical judgment still matter. Those are learnable skills and real constraints, not reasons to stay outside.
Knowing AI trivia is less useful than building a repeatable habit of turning ideas into tested outcomes. Speed helps only when the result is useful, responsible, and maintainable.
Hype vs. leverage¶
Not everything with "AI" on the label deserves your time. A working filter:
| Hype | Leverage |
|---|---|
| "AI will replace everyone" | "Which bounded task should I time before and after a controlled trial?" |
| Buying courses about prompts | Shipping one real project this month |
| Arguing about AGI on the internet | Automating your own Monday morning |
| Waiting for the tools to "settle down" | Learning the concepts that survive every tool change |
This course focuses on the right column. The concepts you'll learn — the agentic loop, context engineering, verification, and orchestration — are candidates for transfer because they concern how you direct machine work. You still need to test their implementation and limits in each tool and domain.
The honest caveats¶
Real leverage requires real honesty:
- Models confidently make things up (Chapter 2 covers why, and what to do about it).
- Agents can act wrongly as easily as they act rightly — which is why verify is a permanent step in your loop, never an optional one.
- The tools change fast. Specification, verification, and risk-management practices are more portable, but their implementation still needs retesting in each tool.
Try it now¶
No terminal yet — just paper (or your notes app):
- Write down the three tasks that eat most of your week. Be specific: "Monday status report," not "admin."
- For each, record the current time and quality baseline, then define what improvement would justify a trial.
- Write down one thing you've wanted to build or launch but didn't, because "I'm not technical" or "I can't afford a developer."
Keep this list. In Chapter 6, you'll start crossing things off it — for real.
Check your understanding¶
- What are the three shifts that made this AI moment different from past hype cycles?
- What is the "gate" for this technology wave, and why is it different from the gates on the internet and mobile waves?
- What's the difference between AI hype and AI leverage? Give one example of each from your own life.
- Why is verification a permanent part of working with AI rather than a beginner phase you grow out of?