Explorations Course

The opening session establishes what data science is, what AI has changed about it, and why neither is the exclusive province of people who write code. Participants encounter the vocabulary and the landscape: what counts as data, where it comes from, what kinds of questions data can and cannot answer, and how AI systems have altered who is able to ask them. The session sets the premise the rest of the course rests on: that domain knowledge is not a deficit to be corrected before technical work can begin, but the thing that makes technical work worth doing. Participants leave with a working map of the field and a sense of where their own expertise sits on it.

Experiential Learning Studio: No-coding tools for data and AI analytics. Hands-on introduction to the intuitive, plug-and-play platforms used throughout the course, so that participants can begin working with real data from the first weeks without a programming prerequisite.

The first of three design thinking sessions introduces a human-centered approach to innovation and problem-solving. Design thinking begins not with a solution but with a rich understanding of the messy reality of people’s lives, with their beliefs, values, motivations, and aspirations, and treats that understanding as the raw material of good problem definition. The session covers the history and evolution of the approach and walks through its five stages: empathy, problem framing, ideation, prototyping, and testing. For participants who will spend the year working on real problems for real organizations, this is the session that establishes how to tell a well-posed problem from a poorly posed one.

Experiential Learning Studio: Framing a problem statement. Participants translate a domain concern into a problem statement that data can actually address, and form project groups.

Before any model is built, someone has to look at the data. This session covers what that looking consists of: understanding structure and distribution, identifying what is missing and what that absence might mean, spotting outliers and deciding whether they are errors or the most interesting thing in the dataset, and forming initial hypotheses that later analysis can test. Emphasis falls on honest practice — documenting what was excluded and why, resisting the temptation to see a pattern that is not there, and recognizing when the data cannot support the question being asked of it. Participants work with real, imperfect datasets rather than teaching examples.

Experiential Learning Studio: Group work on a real-life dataset. Teams work through a genuine, imperfect dataset: examining structure and distribution, identifying what is missing and what that absence implies, and reporting what the data can and cannot support.

Bias enters analytical work through two distinct doors, and this workshop treats them together because in practice they compound. The first is psychological: the cognitive foundations that shape how people perceive, categorize, and judge before any data is involved. The second is systemic: how societal bias becomes algorithmic bias when historical patterns are encoded into training data and model design. Participants work through strategies for recognizing and countering their own biases, methods for detecting and measuring bias in data and model outputs, and what it takes to move toward fairer AI systems. A case study on biased outcomes in automated hiring tools anchors the abstractions in a setting participants recognize: one where the harm is concrete, the mechanism is traceable, and the people affected rarely know it happened.

Experiential Learning Studio: Auditing a dataset for bias. Working in teams, participants examine a real dataset and its documentation for what was collected, what was omitted, how categories were defined, and whose experience the data does and does not represent, then draft the disclosure they would attach to any analysis built on it.

Built directly on the article by Dunning, Heath, and Suls titled “Flawed Self-Assessment: Implications for Health, Education, and the Workplace,” this workshop examines a well-documented finding with uncomfortable implications: people are poor judges of their own competence, correlations between self-assessment and objective performance are modest at best, and others often judge us more accurately than we judge ourselves. Participants work through the meta-analytic evidence, the above-average effect, overconfidence in prediction, and the double curse of incompetence, or the difficulty that the skills needed to perform well are often the same skills needed to recognize that one is performing badly. The material is placed early in the program deliberately. Participants are about to spend a year producing analyses in domains where they are new, using tools that make confident-looking output easy to generate. The session gives them a vocabulary for their own uncertainty and a set of habits, including seeking disconfirming evidence, building in external checks, calibrating claims to what the evidence actually supports, for working responsibly at the edge of their competence.

Experiential Learning Studio: Calibration in practice. Participants estimate their own confidence on a set of judgment tasks, compare their estimates against outcomes, and discuss where their calibration was off and why, then identify the checkpoints in their own capstone work where an external check would catch what self-assessment would miss.

A finding that cannot be communicated has no effect. This session covers the construction of clear, honest visualizations and the narrative work of turning analysis into something an audience can act on. Participants learn the choices that shape how a chart is read, including scale, framing, color, what is included and what is cropped out, and how the same data can be made to tell very different stories, which is as much a lesson in reading visualizations skeptically as in making them. The session pays particular attention to communicating uncertainty and limitation without either burying them or letting them swamp the finding.

Experiential Learning Studio: Building interactive dashboards. Participants construct a working dashboard and present it to peers, with feedback focused on whether an audience unfamiliar with the data can read it correctly.

A workshop on articulating professional purpose, delivered in two parts. The first develops a personal mission statement: what it does, whether you need one, and how to write one that survives contact with reality. Participants work from three tests of a meaningful mission: that it be understandable to someone outside your own context, that it be self-igniting rather than obligatory, and that it align with genuine values and competencies rather than aspirational language. Guided questions and worked examples move participants from abstraction to a statement they can actually stand behind. The second part turns to the personal tagline: a short, concrete articulation of the value you offer, drawn from your core strengths, framed around value rather than tasks, and written in active language. Supporting readings include Tom Peters’s “The Brand Called You” and materials on resume writing as a form of professional self-definition. For participants navigating a transition, such as students entering a job market reshaped by AI or professionals repositioning within their organizations, this is the session that asks what, specifically, they are now able to offer.

Experiential Learning Studio: Tagline workshop. A six-step in-class sequence: strength discovery, value translation, audience targeting, drafting, structured peer feedback, and refinement. Participants leave with a tagline tested against an audience rather than written in isolation, and a reflection posted to their program journal.

The session that turns the year’s learning into a deliverable. Participants receive the capstone requirements in full: a clearly articulated problem or question; a justification for why AI or data tools are the right approach to it; a description of data sources, whether real or simulated; an analysis or prototype demonstrating AI-supported reasoning; explicit treatment of ethical, social, and governance considerations; and a reflection on limitations, risks, and what would need to be true for the work to go further. The emphasis throughout is that a capstone is not a demonstration of tool proficiency but an argument about a real problem, with the analysis serving the argument. The session also covers internship and RAISEuP fellowship placement: how matching works, what host organizations expect, and how capstone domain choices shape placement options. Teams leave with a submission deadline and a first deliverable.

Experiential Learning Studio: Project scoping and prototyping. Teams draft their project title, problem statement, and data plan, and build a first analysis prototype in conversation with an AI assistant, establishing early whether the question they have chosen is one their data can answer.

This session opens the black box. Participants learn what a model is, how learning from data actually proceeds, what distinguishes supervised from unsupervised approaches, and where large language models fit within the broader field. The goal is not to build models from scratch but to understand their mechanics well enough to judge their output: what a model is doing when it makes a prediction, what it cannot know, how it fails, and why fluent output is not the same as correct output. Participants who arrive thinking of AI as a sophisticated search engine typically leave thinking of it as something with mechanisms that can be reasoned about.

Experiential Learning Studio: Working with large language models. Guided practice in directing an AI assistant through an analytical task, evaluating its output, and identifying where it went wrong.

Statistical modeling, taught from the question rather than the equation. The session defines a model in human terms, such as a simplified representation of complex real-world patterns, a way to estimate relationships between variables, a method for summarizing evidence, and insists on two things about it: a model is not a statement of truth but of support, and it is always shaped by human choices and assumptions. Participants learn to translate a human question into a modeling question, to think in variables without writing code, and to work through a model in plain language: describing the dataset, specifying outcomes and predictors, asking an AI assistant to recommend an appropriate approach, requesting explanations alongside results, and treating what comes back as a draft rather than a final answer. The session is explicit about the role AI plays here: a research assistant rather than a decision-maker, automating technical execution while remaining dependent on human framing, and capable of amplifying insight and error alike. It closes on limits and responsibility: all models rest on assumptions, data can encode historical bias and inequality, models mislead when overinterpreted, ethical risk rises as tools become easier to use, and human accountability does not transfer to the system.

Experiential Learning Studio: Selecting a model for your project data. Teams bring their capstone data, work through the specification process with faculty supervisors, and defend their choice of approach — including what the model cannot tell them.

Facilitated by KC7 Foundation Inc.
KC7 redefines cybersecurity fundamentals not as a set of technical skills but as transferable, cross-disciplinary capabilities: critical thinking, teamwork, written and verbal communication, and the application of geopolitical context. Participants investigate a simulated intrusion, querying real-scale data to trace an adversary’s activity and reconstruct what happened — building analytical reasoning through the investigation itself before the technical content arrives. Delivered in two parts, a lecture session followed by a separate guided hands-on session with staff support, the format participants identified as the most effective in the program.