The Concerns Worth Having
The concerns executives actually raise, a response to each, and the module where it is answered in full. Read it once before Week 1.
Time to read: about 3 minutes
Most people arrive carrying a version of one of these. Every one is reasonable, and each points at something the program is built to answer. Read the headlines first; the one that is most yours is the one to bring to the peer share.
"It will take my job."
AI replaces tasks, not roles. What is emerging is amplification, and the gap that is widening is between the AI-fluent and the AI-avoidant. Your judgment and your accumulated context are the half a model does not have.
Answered across the Dialectic, Modules 2 to 4, where you direct the work at every step.
"It makes things up."
True, and it is the most important thing to understand about these tools. A model predicts plausible text, it does not retrieve facts, so hallucination is the architecture rather than a bug awaiting a patch. The answer is to build verification into the workflow.
Answered in Module 3, Working by Delegation, on reviewing output produced without being watched.
"I'll become dependent and lose my edge."
It turns on one distinction: are you outsourcing your thinking or augmenting it? One well-directed hour can stand in for ten, but you still owe two hours of review, and that review is where your judgment stays sharp. You decide what to hand off and what to keep, on purpose.
Answered in Module 3 and Module 4, on where the trust boundary sits and when to keep your hands on the work.
"My data isn't safe."
Some tools train on your conversations by default and some do not, so know your tool's policy and keep raw client data out of consumer tiers. Beyond the tool, safety is governance: what the system may touch, and what your organization commits to in writing.
Answered in Module 6, Governance and Enablement, where you draft a first AI policy for your team.
"It just tells me what I want to hear."
Models lean agreeable, which is comfortable and is not the same as being right. The working defense is to make the model argue the other side, and to cross-check anything that matters against more than one model.
Answered in Module 2 and its reading The Winter Method, then again in Module 7 when you compare models instead of trusting one.
"There's bias baked in."
A model learns from human-generated text, and that text carries human bias, which surfaces in what it treats as the default and whose perspective it centers. The move is not to pretend it is neutral; it is to apply your judgment, most of all where the stakes are high: in hiring, evaluation, and anything customer-facing.
Answered in Module 5, Transparency and Disclosure, on standing behind AI-assisted work.
"My people are skeptical, and there are too many tools."
You will not convince a skeptic with enthusiasm; you convince them the day you solve a real problem faster than they thought possible. The tool sprawl is why the program teaches the four Ds, a fluency that carries from one product to the next.
Answered in Module 6, on leading adoption, and Module 7, on choosing what to adopt.
Beyond the headlines
Context is lost between sessions. A model forgets you when the window closes, which is the whole reason the five Elements exist. Module 1.
The verification burden. AI shifts work from creating to checking, and checking is not always faster. Sometimes the human-only path still wins. Module 3.
Rabbit holes and lost time. It is easy to spend three hours on a thirty-minute task, so set a time box before you start. A habit to build from Week 1.
Disclosure. Knowing when to say AI did the work, especially in regulated and public-facing settings. Module 5.