Kick-off

Research Project 1 (RP1) — U1

Andy Weeger

Neu-Ulm University of Applied Sciences

September 3, 2026

Opening Remarks

Warm-Up

Art?
Galaxy?
Andromeda?

It is a galaxy, the Andromeda Galaxy, to be exact.

• Who told you that?
• What makes up a galaxy?
• How is a galaxy composed?
• What is the power that drives the stars around its center?

These questions, and their answers, originate in research. Research provided the concept “galaxy,” which is very abstract, well outside our direct experience, and used it to educate people, to make them able to interpret the picture.

Closer to your own field: think of a term like “digital transformation” or “agentic AI.” Both are abstract concepts that originated in research and are now used every day by people who never questioned where they came from, or how solid the ground under them actually is. That question, how solid is the ground, is what this module trains you to ask about your own challenge.

Research

Research means that you don’t know, but are willing to find out. Charles F. Kettering, American engineer

That stance, admitting you don’t know, and being willing to systematically find out, is what separates this module from a consulting engagement: your task is to understand a problem you do not yet fully understand, on purpose, before you are allowed to propose anything, rather than to apply a solution you already have.

Learning Outcomes

By the end of this unit, you will be able to:

• Distinguish basic from applied research, and hold your own project to the rigor-relevance standard every deliverable this semester is judged against.
• Tolerate not-yet-understanding a problem, and defend a decision to keep investigating rather than jump to a fix, despite incomplete information.
• Identify what input or expertise you need next, and use push units, pull sessions, and coaching to get it.

Research Project

Researcher Role

This semester, your job is to find and understand a problem well enough that a solution could be judged against it, not to deliver one yourself.

Unlike an industry project team, which divides tasks, moves fast, and converges on a recommendation by week three, a researcher team holds a question open longer than feels comfortable, treats disagreement between sources as data rather than a problem to smooth over, and stays accountable to evidence rather than a deadline-driven consensus.

The research project 1 module is deliberately structured to slow down the jump to solutions as the recommendation is only as good as the problem understanding underneath it.

GenAI Use

GenAI tools are allowed.
Undisclosed use is not.

Every paper you submit this semester (Problem Framing Brief, Review Protocol, Paper) carries a mandatory declaration: which tools, where, and how you checked their output.

The rule is simple: use GenAI tools the way you would use any other tool, provided you can stand behind the result. The declaration names which tool you used, for which section, for what purpose, and how you checked its output, and every template already has the table for it. Leaving it blank while having used a tool falls under the same “good scientific practice” clause that covers plagiarism: automatic failure of that deliverable, not a deduction.

The declaration also makes you responsible for what you submit. If a tool invents a citation, states something false, or misreads a source, that error is yours once you have adapted and submitted the text, not the tool’s.

GenAI Caveats

• Judgment erosion: offloading a task to AI speeds it up today and measurably weakens your own unaided performance on it later (Bastani et al., 2025; Budzyń et al., 2025).
• Silent errors: a GenAI-drafted screening call, extraction, or synthesis sentence looks as plausible as a careful one, and catching what’s wrong in it takes as long as doing the reading yourself (He, Agarwal, et al., 2026; He, Miller, et al., 2026).
• Homogenization: teams that prompt the same model the same way tend to converge on the same familiar framing, working against the novelty a good research topic needs (Doshi & Hauser, 2024; Wadinambiarachchi et al., 2024).

Judgment erosion. Bastani et al. (2025) found that high-school students using a generative AI tutor solved practice problems faster, then performed measurably worse on an unaided exam once the AI was removed: the effort that normally builds a skill had been offloaded, so the skill never accumulated. Budzyń et al. (2025) found the same pattern in trained professionals, endoscopists accustomed to continuous AI assistance showed reduced unaided detection accuracy once it was taken away. A competence check-in asks what you can now do that you could not do before, and offloaded judgment leaves nothing there to report.

Silent errors. He, Agarwal, et al. (2026) name the gap directly: generation now happens at machine speed, but verification, the reading, tracing, and checking needed to catch a subtle error, still happens at human speed. He, Miller, et al. (2026) confirm the downstream cost in a longitudinal study of real software projects: AI-assisted work raises short-term output while measurably increasing long-term complexity and error, because work that looks right on a first pass can carry flaws nobody caught. Applied to your own review, a screening decision, coded extraction, or concept-matrix cell drafted by AI is only as trustworthy as the verification you actually did on it, and an unverified shortcut here does not stay contained to one deliverable, it becomes a silent error in the corpus the rest of your synthesis, and your paper’s conclusions, are built on.

Homogenization. Doshi & Hauser (2024) found that while an individual using an LLM produces more ideas, a population of people prompting the same model converges on more similar outputs to each other than an unaided population does. Wadinambiarachchi et al. (2024) trace the mechanism to design fixation: early exposure to an AI-generated framing anchors a person’s own thinking even when they are told to work independently. U4: Effective Topics already asks you to argue your topic against the familiarity, maturity, and nearness traps; the same risk applies to how you frame it. A Problem Framing Brief built from a model’s first suggestion risks the same familiar angle every other team asked for, instead of the tension your own map of perspectives actually surfaced.

Research Stance

Types of Research

Basic research
& applied research

Basic research is intended to enhance basic knowledge about the physical, biological, psychological, or social world, or to enlighten historical, cultural, or aesthetic phenomena. It is not goal-directed.

Applied research is intended to address issues immediately relevant to current practices, procedures, and policies, and to improve human decision-making. It aims to intervene in or alter a phenomenon, and is, thus, goal-directed (Recker, 2021).

Your module sits deliberately between the two. It has an applied trigger, a real industry challenge, but Phase 1 through 3 ask you to treat it with the patience of basic research: describe the phenomenon carefully, before you let the goal-directedness of “applied” pull you toward a fix.

Discussion

Turn to your neighbor. Locate your own first hunch about the challenge on the basic-applied spectrum you just saw.

State which pole it sits closer to, and why, in one sentence each.

05:00

Definition

Research includes in-depth investigation or testing, especially critical and exhaustive investigation or experimentation, aimed at the discovery of new facts and their correct interpretation, the verification of accepted conclusions, theories, or laws in light of newly discovered facts, or the practical application of such conclusions, theories, or laws (Recker, 2021).

What research is about:
With rigor acquired relevant knowledge

In a nutshell:

• If there is no rigorous, critical, and exhaustive investigation, there is no research.
• If there is no discovery, there is no research.
• If there is no interpretation of the data, there is no research.
• If there is no formation, revision, or application of theories, there is no research.

Rigor and Relevance

Rigor

The degree to which research methods are methodically and accurately carried out, to identify important influences occurring in the process of conducting the research. It is a set of standards investigators use to evaluate the quality, trustworthiness, and value of research.

Relevance

The degree to which findings of the research are relevant to and can be applied to a real-life situation.

Hold both words in mind for the rest of the semester. Every deliverable in this module, the Map of Perspectives, the Problem Framing Brief, the Review Protocol, the final paper and pitch, is judged against both. A rigorous review of an irrelevant question wastes effort. A relevant recommendation built on a sloppy review cannot be trusted. Every deliverable in this module is judged on both criteria at once.

Reflection

Write down, for yourself, not to hand in:

1. One thing about the semester ahead you are genuinely curious about.
2. One thing about it that makes you nervous.

This is your first research log entry. Keep both.

03:00

Homework

Open your research log today. Write your first entry before the next session.

Your positioning statement, half a page:

1. Which aspects of the challenge do you, personally, already have a stake in, prior experience, a strong opinion, a professional connection?
2. What do you want to be able to do by January that you cannot do today?

Keep the research log running for the rest of the semester. You will draw on it directly for later competence check-ins.

Q&A