Most people who use generative AI tools daily have little idea what is actually happening under the hood. This workshop addresses that gap directly, without requiring a mathematics or computer science background.
What the day covers
The morning session focuses on how language models are trained, what tokenization means in practice, and why the same prompt can produce different outputs each time. These are not abstract concepts here — participants run live experiments with real models and observe the differences firsthand.
The afternoon shifts to applied work. Participants bring their own use cases: drafting documents, summarizing reports, writing code snippets, or generating structured data. Each scenario is tested, critiqued, and refined in small groups.
Where models tend to go wrong
Hallucination, bias in training data, and context window limitations are examined through specific examples rather than general warnings. Participants learn to spot outputs that look confident but contain factual errors — a skill that takes practice to develop.
One participant, a legal analyst named Oksana Dreher, noted that she had been using an AI tool for six months before realizing it was silently omitting key clauses from contract summaries.
Workshop participant feedback, March 2024
Who attends
The workshop draws professionals from finance, communications, operations, and research. No prior AI experience is required, but basic computer literacy is assumed. Groups are kept small — twelve participants maximum — so discussion stays substantive.