Drafting
Produces a first version quickly.
Module 1
What AI Actually Is
AI can be useful without being reliable. Fluency is not understanding, and confidence is not correctness.
AI systems generate outputs by learning patterns from vast amounts of training data. You provide a prompt - a question, a task, an instruction - and the system responds with text that matches the pattern of a correct-looking answer.
A video introduction is planned for a future update.
I built an AI-assisted trading platform with no prior AI experience.
I assumed that if the output looked right, it probably was. I was wrong. What the AI produced was plausible - statistically consistent with what a correct answer looks like - but not reliably accurate. I had to build verification layers I had not planned for, slow down a system I had designed to move fast, and retest things I had already considered done. This guide is what I wish I had known before I started.
How AI Systems Work
These systems can produce genuinely useful output: language, code, summaries, classifications, and analysis. What they cannot do is understand what they are producing. A well-formed, polished answer is not evidence the system verified anything.
Hallucination: when an AI generates something that sounds correct but is factually wrong. It is not intent. It is not deception. It is a structural property of how these systems work. It cannot be eliminated, only managed.
In 2023, attorneys in Mata v. Avianca, Inc. (S.D.N.Y.) submitted a legal brief containing AI-generated case citations. The cited cases did not exist. The court described them as “bogus” and sanctioned the attorneys. The output was polished and looked authoritative. Hallucination does not announce itself.
What AI Is Good For
Summarizing
Compresses large amounts of text.
Rewriting
Changes tone, structure, or clarity.
Brainstorming
Generates options and angles.
Transformation
Converts styles, structures, and formats.
Pattern assistance
Helps surface possible themes or inconsistencies.
What AI Is Weak or Risky For
| Risky Use | Why It Needs Control |
|---|---|
| Factual claims | May fabricate or misstate, confidently. |
| Legal, medical, financial, or HR tasks | Stakes are high; context matters; AI cannot take responsibility. |
| Arithmetic and formal logic | May appear reasoned while making basic errors. |
| Security-sensitive tasks | Can leak, mishandle, or be manipulated. |
| Autonomous action | Can act on bad assumptions at machine speed. |
| Long planning chains | Can drift, omit constraints, or contradict itself. |
Is AI Appropriate for This Task?
- Does this require factual accuracy? Verify independently before using the output.
- Does it involve sensitive or private data? Do not paste until organizational policy is clear.
- Could someone be harmed by a wrong answer? Require human review before action.
- Could money, access, employment, safety, or legal status be affected? Treat as High or Critical risk.
- Will AI take action automatically? Require formal controls before proceeding.
- Would you be comfortable explaining this decision later? If no, stop.
Get Notified of Updates
This guide is updated when vendor policies, regulatory guidance, or risk patterns change. Enter your email to be notified — no sales pitches, no newsletters.