Saginaw Valley State UniversityZahnow Library
Imagination LabAPPLIED AI LITERACY · A LIBRARY LEARNING GUIDE
Zahnow Library guides/Applied AI literacy
LEARN · EXPERIMENT · REFLECT

Applied AI lessons

10 lessons and six hands-on experiments. Work through them in order, at your own pace.

Lesson 8 of 10

Design thinking, building, and the building loop

In plain termsDesign thinking is a way to solve problems by starting with people, not technology: understand who you’re helping, define their problem, brainstorm, build a quick prototype, and test it. AI makes building fast — which makes talking to real people first even more important, so you don’t quickly build the wrong thing.

Design thinking and user research

Turn imagination into things people want

Design thinking is a human-centered way to solve problems. Test sends you back to Empathize.

  1. 1
    Empathize

    Understand people.

  2. 2
    Define

    Name the problem.

  3. 3
    Ideate

    Generate many ideas.

  4. 4
    Prototype

    Make it fast.

  5. 5
    Test

    Try it and learn.

AI speeds up every step except caring about the person you’re building for.

Know your user: the research methods that drive iteration

User interviews

Uncovers motivations, frustrations, and mental models.

Usability testing

Watch real users attempt real tasks to reveal friction.

Surveys & questionnaires

Quantify attitudes and behaviors at scale.

Journey mapping

Visualize the full experience, from first touch to final outcome.

A/B testing

Let real behavior decide between two versions.

Analytics & heatmaps

See where users go, click, and drop off.

Takeaway: Great products are built on evidence. AI can now assist with every method, from synthesizing transcripts to surfacing patterns in data.

Sources: Stanford d.school; Simon (1969); IDEO.

In plain termsA startup once needed designers, coders, marketers, writers, and more. Today, one motivated person with AI can cover a first version of many of those roles. That doesn’t make experts unnecessary, but it means you can test an idea before you have a team, money, or permission.

Everyone becomes a builder

Everyone it takes to run a startup

A traditional startup team vs. what one founder with AI can cover today. (Illustrative role list, not any specific company’s org chart.)

Founder / CEO

Human accountable

Vision and decisions: stays human.

Product manager

AI can draft

Specs and roadmaps: ChatGPT, Claude.

Software engineers

AI can draft

Code: Claude, Codex, Lovable.

Designer

AI can draft

UI and brand: Midjourney, Canva.

Marketer

AI can draft

Copy and ads: ChatGPT, Runway.

Sales

AI can draft

Outreach and proposals.

Customer support

AI can draft

Chatbots and voice agents like ElevenLabs.

Data analyst

AI can draft

Dashboards and reports.

AI can fill the seats. Someone still has to own the decisions. Medvi outsourced the regulated parts (doctors, pharmacy, compliance), and that is where the trouble showed up (as reported).

Sam Altman’s bet: one person, one billion dollars

The bet

In 2024, Altman said AI would make a one-person billion-dollar company (a “unicorn”) possible, something that used to take a whole team.

Some say it already happened

Medvi, a telehealth company led by Matthew Gallagher with AI tools, reported $401M in 2025 sales. Some people point to it as the first example.

Why it is profound

The hard part used to be hiring and paying a team. If AI does that work, the scarce thing is the idea and the person with the will to build it.

One person. Superpowers.

  1. 1
    Make anything

    A song, a video, and a logo before lunch.

  2. 2
    Build big

    A working app for your club, no dev team.

  3. 3
    See far

    Spend your time deciding what is worth making.

Your superpower is direction. AI does the heavy lifting.

Sources: Fortune 2024; NYT 2026.

In plain termsAs AI does more of the hands-on work, your role shifts from writing every word yourself (author) to setting direction, standards, and the final vision (architect). It’s similar to being an editor or a director: you decide what the project is for and what “good” means.

From author to architect

Ascend the hierarchy: orchestrate AI toward one vision

AI lets us move from individual creators to editorial directors who orchestrate AI agents toward one unified vision. The top tier sets the course; the bottom tier does the work.

  1. 1
    Editorial Director

    Sets the overarching vision, strategy, and creative direction.

  2. 2
    Managing Editor

    Streamlines workflows and coordinates content for timely delivery.

  3. 3
    Section Editor

    Curates content within a domain, ensuring quality and coherence.

  4. 4
    Writer

    Generates individual pieces and crafts compelling narratives.

Takeaway: With AI, we can embody all these roles at once, guiding our digital workforce toward a single creative goal.

In plain terms“Vibe coding” means building software by describing what you want in plain language and letting an AI write the code. You don’t need to be a programmer to try it, but you do need a loop: describe, let it build, test it yourself, then steer with specific feedback. Small steps beat giant requests.

Codex and vibe coding

A technical teammate, and a 5-step loop for building

Codex is OpenAI’s coding agent: describe what you want, and it writes, tests, and fixes the code. Describe it → Codex builds it → you test and steer.

What it could help you build or fix

A personal portfolio website

Describe the look and sections.

A class project app

A tool, game, or dashboard from a plain-language idea.

Messy data, cleaned

Scripts that clean and chart a dataset.

Code you do not understand

Explain it, find bugs, suggest fixes.

How to vibe code: a 5-step loop

Repeat
1Describe itSay what you want in plain words.
2Build a first versionLet the AI write it.
3Run and lookClick around. What works? What breaks?
4Fix with feedbackPaste the error or say what to change.
5Save and repeatKeep versions. Make small changes.

Tools: Lovable, Replit, Cursor, ChatGPT Codex. Never paste passwords or private data.

Try it

Build me a simple website that helps SVSU students find study spots. Make it clean and mobile-friendly.

Takeaway: You do not need to write the code. You do need to decide what is worth building — and judge the result. Always test: Codex can make mistakes and use real tools.

Sources: OpenAI Help Center; openai.com/codex.

Hands building a small house from blocks, with blueprints and a crane

Check your understanding

1. You have an app idea for campus. What’s the design-thinking first step?
2. “From author to architect” means your role shifts toward…
PreviousNext: experiment