AI-R01 — AI-Assisted Research Design

Course Code: AI-R01  |  School: School of Artificial Intelligence  |  Cluster: Level 2 — Research & Academic Practice

Level: Intermediate  |  Duration: 6 weeks · 20–26 learning hours  |  Language: English  |  Certificate: Professional Certificate (non-degree)  |  Format: Self-paced with AI support under human supervision

Overview

Research design is the stage at which AI assistance is most tempting and least appropriate. A generative system will produce a plausible research question, a plausible framework and a plausible method in seconds, and every one of them will be a composite of the field’s conventional wisdom rather than a response to a real gap.

This course teaches design as a sequence of defensible commitments: a problem worth studying, a question answerable with available means, a framework that constrains rather than decorates, a design that matches the inferential claim, and a plan for the data that will actually be obtainable. AI is used deliberately and narrowly within that sequence, as a device for generating alternatives to be criticised and for stress-testing a design against objections the researcher has not thought of.

The final artefact is a complete research protocol of the kind that would go to an ethics committee or a funder, accompanied by a documented record of where machine assistance was used and what it got wrong.

Learning outcomes

On completion, a successful learner will be able to:

1. Convert a topic of interest into a researchable question with a stated unit of analysis and a specified claim type.
2. Justify a research design against the inference you intend to draw, and state what the design cannot support.
3. Use a theoretical framework to constrain measurement and interpretation rather than to decorate the introduction.
4. Plan data collection that is feasible given real access, ethics and time constraints, with a documented fallback.
5. Employ AI assistance for alternative generation and adversarial critique, while retaining authorship of the design.
6. Anticipate ethical review, including consent, data protection and the disclosure of machine assistance.
7. Produce a protocol another researcher could execute and critique.

Who this course is for

Doctoral and master’s researchers, early-career academics, institutional researchers and analysts who design studies, and supervisors who want a common vocabulary for design critique.

Prerequisites

Prior exposure to research methods at postgraduate level, or substantial applied research experience. AI-F01 recommended. Learners must bring a real research problem they intend to pursue.

Syllabus

Module 1 — From topic to question

Focus. The difference between an interest, a topic, a problem and a question. Specifying unit of analysis, scope conditions and the type of claim being made: descriptive, associational, causal or interpretive.

Lessons. 1.1 Interest, topic, problem, question. 1.2 Unit of analysis and scope conditions. 1.3 Claim types and their evidentiary requirements. 1.4 Generating and discarding candidate questions.

Core reading. Wayne C. Booth, Gregory G. Colomb & Joseph M. Williams, The Craft of Research, 4th edition (Chicago: University of Chicago Press, 2016), chapters 3–4.

Deliverable. Question specification: one question with unit of analysis, scope conditions and claim type, plus three rejected alternatives with reasons.

Module 2 — Framework as constraint

Focus. Choosing a theoretical framework because it generates testable expectations, not because it is fashionable in the field. Deriving observable implications and identifying what would count as disconfirmation.

Lessons. 2.1 What a framework is for. 2.2 Deriving observable implications. 2.3 Rival explanations. 2.4 Stating disconfirming evidence in advance.

Core reading. John W. Creswell & J. David Creswell, Research Design: Qualitative, Quantitative, and Mixed Methods Approaches, 5th edition (Thousand Oaks: SAGE, 2018), chapters 3 and 7.

Deliverable. Framework memo with derived expectations and stated disconfirming conditions.

Module 3 — Matching design to inference

Focus. Experimental, quasi-experimental, comparative, case-based and interpretive designs, and the specific inferential claim each supports. Threats to validity and the design choices that address them.

Lessons. 3.1 Design families and their claims. 3.2 Internal validity and its threats. 3.3 External validity and the transportability question. 3.4 Case selection logic.

Core reading. William R. Shadish, Thomas D. Cook & Donald T. Campbell, Experimental and Quasi-Experimental Designs for Generalized Causal Inference (Boston: Houghton Mifflin, 2002), chapters 1–3. John Gerring, Case Study Research: Principles and Practices, 2nd edition (Cambridge: Cambridge University Press, 2017), chapters 2 and 5.

Deliverable. Design justification with a validity threat table and the mitigation for each threat.

Module 4 — Using AI without ceding authorship

Focus. Three legitimate uses: generating alternatives for the researcher to reject, articulating counterarguments, and checking a design against a documented reporting standard. Three illegitimate uses, and how to recognise the drift from the first to the second.

Lessons. 4.1 Alternative generation and disciplined rejection. 4.2 Adversarial critique of your own design. 4.3 Checking against reporting standards. 4.4 Where assistance becomes substitution.

Core reading. Gerit Wagner, Roman Lukyanenko & Guy Paré, “Artificial Intelligence and the Conduct of Literature Reviews”, Journal of Information Technology 37, no. 2 (2022): 209–226.

Deliverable. AI-use log: a documented record of assistance, including at least three instances where the machine output was wrong and how you established that.

Module 5 — Data, access and feasibility

Focus. The plan that fails is usually the plan that assumed access. Sampling and case selection, instrument design, gatekeepers, timelines, and a documented fallback for the most likely access failure.

Lessons. 5.1 Sampling and selection in practice. 5.2 Instruments and pilot testing. 5.3 Gatekeepers, permissions and realistic timelines. 5.4 Fallback design.

Core reading. Creswell & Creswell, Research Design, chapters 8–9.

Deliverable. Data plan with timeline, access strategy and documented fallback.

Module 6 — Ethics, protocol and pre-registration

Focus. Assembling the protocol. Consent and data protection, handling of sensitive material, disclosure of machine assistance, and the case for pre-registering analytic commitments where the design permits.

Lessons. 6.1 Consent, harm and data protection. 6.2 Sensitive data and secure handling. 6.3 Disclosure of AI assistance in protocols and publications. 6.4 Pre-registration: what it fixes and what it cannot.

Core reading. Brian A. Nosek et al., “The Preregistration Revolution”, Proceedings of the National Academy of Sciences 115, no. 11 (2018): 2600–2606.

Deliverable. Final submission: complete research protocol with ethics section, AI disclosure statement and, where applicable, a pre-registration draft.

Assessment

Pass mark 70 per cent. All assessed components must be attempted. Every mark in this course is issued by a human assessor; no assessment outcome is generated automatically.

Rubric criteria

Each assessed artefact is marked against four criteria at four levels (distinction, pass with merit, pass, fail).

1. Precision of claim: is it clear exactly what the study will and will not be able to conclude?
2. Design–inference fit: does the chosen design actually support the intended claim, with threats identified and addressed?
3. Feasibility: is the plan executable with the access, time and resources the learner genuinely has?
4. Authorship: is the intellectual content of the design demonstrably the learner’s, with machine assistance disclosed and critically handled?

Reading list

Core. Wayne C. Booth, Gregory G. Colomb & Joseph M. Williams, The Craft of Research, 4th edition (Chicago: University of Chicago Press, 2016). John W. Creswell & J. David Creswell, Research Design, 5th edition (Thousand Oaks: SAGE, 2018).

Design and inference. William R. Shadish, Thomas D. Cook & Donald T. Campbell, Experimental and Quasi-Experimental Designs for Generalized Causal Inference (Boston: Houghton Mifflin, 2002). John Gerring, Case Study Research: Principles and Practices, 2nd edition (Cambridge: Cambridge University Press, 2017).

Peer-reviewed. Brian A. Nosek et al., “The Preregistration Revolution”, PNAS 115, no. 11 (2018). Gerit Wagner, Roman Lukyanenko & Guy Paré, “Artificial Intelligence and the Conduct of Literature Reviews”, Journal of Information Technology 37, no. 2 (2022).

All items are published works identifiable by author, title and publisher. Learners obtain them through an institutional library or the publisher. The Academy does not distribute copyrighted texts.

Academic integrity and use of AI

Generative tools may be used in producing assessed work under three conditions. Use must be disclosed in a short statement appended to each submission, naming the tool and the task it performed. Any factual or technical claim originating from a generative tool must be verified against a citable source before it enters assessed work, and the verification must be evidenced. The analytical judgement in each artefact must be the learner’s own and must be defensible in a short follow-up. The AI-use log is itself assessed. A log reporting no errors in machine output will be treated as evidence that verification was not performed.

Instructor: pending owner confirmation. Pricing: pending owner approval. Reference list verified against publisher records; any later addition is marked for verification before publication.