Module 2
A map for decision-makers. What AI is, where it is going, who is building it, and how to think about your role in it.
Not all AI is the same. Understanding the distinctions helps you ask better questions about risk, oversight, and where your judgment is required.
Generative AI does not retrieve information. It generates it - predicting what words, images, or code should follow based on patterns learned from training data. That distinction matters for how you use it.
When you ask Claude a question, it is not looking up an answer. It is generating the most plausible continuation of your prompt based on patterns in its training. This is why it can be wrong with complete confidence.
The quality of what generative AI produces is directly tied to the quality of what you give it. Vague prompts produce vague outputs. Specific context - your role, your goal, your constraints - produces useful work.
Generative AI produces drafts, analysis, and options. Your job shifts from creation to evaluation. That is not a reduction in your role - it is a reorientation of where your judgment is applied. In practice, semi-automated workflows like auto-drafted emails or suggested replies already blur the line between generative and agentic - even when marketed as simple AI features.
Generative AI is a conversation. Agentic AI is a delegation. The difference determines where your oversight is required - and what happens if you skip it.
AI produces. You decide what to do with it.
AI plans and takes action. You define the boundaries.
Four organizations are producing the models that matter for executive work right now. They have meaningfully different philosophies, privacy postures, and strengths. Knowing the field helps you make informed choices about what to use and when.
Anthropic was founded in 2021 by former OpenAI researchers who wanted to build AI differently. The company's defining bet: safety and capability are not in tension. You can build a more reliable tool by being more deliberate about how it is trained.
Most AI models are trained to avoid harmful outputs by showing them examples of what not to do. Anthropic's approach - Constitutional AI - trains Claude using a written set of principles. Claude learns to evaluate its own outputs against those principles before responding. The result is behavior that is more consistent, more predictable, and less prone to being manipulated into producing things it should not. For executives working on sensitive business problems, that consistency is practically useful, not just philosophically interesting.
By default, Anthropic does not use your conversations to train future models. You can share business context, work through sensitive problems, and think out loud without that data being harvested.
OpenAI's consumer product uses conversations for training by default. You must navigate settings to disable this. The Enterprise version operates differently - check your organization's subscription.
This is a landscape orientation, not a technical reference. The distinctions drawn here - particularly between generative and agentic AI - are simplified to be useful for decision-making, not precise for engineering purposes. The model landscape is accurate as of early 2026 and will change. The external links point to primary sources where you can find more current and more precise information. When in doubt about any claim here, follow the link. If a resource link is not loading, try opening it in an incognito window - browser extensions can sometimes trigger security blocks on institutional sites.