FOR060F Practical AI for Doctoral Research

COURSE SYLLABUS (third-cycle)

Course name: Practical AI for Doctoral Research

ECTS/HP: 3 ECTS — 80 working hours required; up to 40 hours optional, practical task with assessed report

Course code: FOE060F

Educational level: Third-cycle course

Eligibility & Entry requirements

Admitted to third‑cycle studies at LTU. If space allows, the examiner may open the course to doctoral students from other universities and, where appropriate, to LTU employees with an approved professional development plan attached to the application.

Prior: completion of LTU’s AI intro course for all staff (LTU001S, 1.5 ECTS) or equivalent familiarity with basic AI concepts, ethics & legislation, and tool overview. Basic knowledge of (at least one): working with Excel, programming in Python, R, or Matlab.

Aim

Equip doctoral students with practical, research‑grade AI workflows that increase productivity while preserving scientific integrity, privacy, and reproducibility. By the end, each participant will have executed a small, new-to-them AI project embedded in their PhD work.

Course content (PhD‑focused, minimal repetition of LTU001S)

• Possibilities & limits in practice — model capabilities vs. failure modes; verification patterns; data privacy & disclosure in research contexts.
• Chat‑based research work — search strategies, literature mapping, reviewer‑style critique, drafting/editing for proposals & papers, and safe citation handling.
• Presentation content preparation — storyboarding talks, figures and slide outlines, speaker notes, and rehearsal checklists supported by AI; peer feedback loops.
• Data analysis & processing (different tracks) — Track 1: data processing opportunities for studies in the humanities (MS Excel/MS Word); Track 2: AI‑assisted coding for data processing in engineering sciences (Python/R/Matlab); Content: EDA, plotting, table/figure generation, and reproducible research (files/notebooks/scripts); reviewing model‑suggested files/code.
• Mini‑project — conceive, build, and document one novel (for the PhD student) AI‑assisted workflow or artifact aligned to the student’s research. Submit a report.

Learning outcomes

• Plan and execute one novel AI‑supported workflow directly relevant to the PhD, with documented prompts, sources, and validation steps. Write a short report on that (LLL).
• Search & synthesize literature with AI tools while auditing claims and managing citations correctly (no fabricated references).
• Draft and revise scientific text (proposal/paper sections, responses to reviewers) using transparent, controllable prompting strategies.
• Build and review an AI‑assisted data analysis or coding task with reproducible outputs (file/scripts/notebooks, versioned data/figures).
• Design content for presentations (structure, figures, slide outlines) and adapt using buddy feedback to improve clarity and impact.
• Evaluate risks & limits (privacy, bias, hallucinations), apply do’s & don’ts for responsible use in academic contexts, and disclose AI assistance.

Course methods

• Short demonstration videos; live workshops; lab assignments; learning‑buddy check‑ins; optional office hours.
• Dedicated live sessions: ask an expert, focus discussion on what is allowed and what not
• In person discussion with PhD students
• Buddy and reflection pattern
• Accessibility: adapted/alternative examination can be provided upon request.

Examination & grading

• Milestone checks during live meetings (micro‑deliverables or demos)
• Mini‑project deliverables (artifact + process log + reflective summary) and a short oral presentation with Q&A spanning over all deliverables.
• Grading scale: Pass/Fail.

Course literature & tools

Curated articles and tool links are provided per module; no single textbook. For chat‑based work we recommend ChatGPT (available under LTU’s license); when doing literature work, use Search and Deep research options; consider image/diagram generation for figures/flowcharts. Excel or programming in Python, R, or Matlab.

Education cycle, frequency, admissions & tuition

• Third‑cycle; offered twice annually (we can start in HS 2025 already).
• Open to external doctoral students if space allows; limit ~40 places.
• If allocated via internal resources, no fee for LTU doctoral students; fees may apply for others per department decision.

Contact & governance

Contact person: Kenneth Paulsen kenneth.paulsen@ltu.se

Examiner: Marcus Liwicki

• Kenneth Paulsen
  Register by writing a mail to Kenneth by February 10, 2026

Indicative schedule (6 live sessions + labs; buddy practice between sessions)

Session Focus Asynchronous Lab

1 — Kick‑off & safe practice

(do and don’ts with AI) Setup, disclosure norms, privacy; failure modes refresher; prompt patterns for controllable outputs; provenance logging; buddy pairing; project ideation. Lab 1: Set up AI stack (accounts, reference manager, repo) and a prompt log template. Reflect: what is OK

2 — Literature & SOTA

mapping (hands‑on) Search → read → verify → cite; claim checking; figure/table extraction. Lab 2: 1–2 page SOTA map using Search/Deep research; export all sources (DOIs/links) + ‘hallucination check’ note.

3 — Writing with AI

(proposals & papers) Outline→draft; tightening methods & results; style control; reviewer‑reply drafting; what AI must not do (e.g., invent citations). Lab 3: 2–3 page draft/revision; color‑code human vs AI edits; run citation audit.

4 — Data analysis & coding

with AI EDA with assistants; unit tests for suggested files/code; plotting; exporting reproducible artifacts. Lab 4: Reproduce a small analysis (own or open data) with an AI‑assisted excel or notebook + tests.

5 — Presentation content

preparation Talk storyboarding; figure briefs; slide outlines; speaker notes; buddy feedback rounds. Lab 5: Slide outline + two figures/briefs; present a 5‑minute research pitch to buddy.

6 — Mini‑project demo day 5–7 minute demo per participant Q&A, and next‑steps plan. Before: mini-project (see below)

Individual exam In-depth Q&A about all

Mini‑project (core assessment)

Goal: Implement one new-to-you AI workflow that measurably helps your PhD. Examples:

• Verified literature review copilot (with claims log and citation audit).
• Reproducible AI‑assisted data analysis (script/notebook + tests).
• Slide/figure content generator (storyboard + two production‑ready figures/diagrams).
• Lab ‘research assistant’ (prompted SOPs, to‑do triage, meeting minutes with action items).

Deliverables (submit 48h before Session 6):

• Artifact (repo, notebook, slides, or toolkit).
• Process log (prompts, versions, sources, validation notes, time required for tasks).
• 2‑page reflective summary (what worked/failed, risks, next steps).
• Short live demo (5–7 min) + Q&A.

Pass criteria (rubric excerpt):

• Novelty & relevance: clearly new to the participant and relevant to their PhD.
• Integrity: transparent AI use; proper source handling; no fabricated citations.
• Reproducibility: others can run or follow the workflow; code/tests where applicable.
• Impact: time/quality gains evidenced (before/after or baseline vs. improved).
• Reflection: risks/limits identified; plan for sustained use.

Do’s & Don’ts for PhD AI work

Do

• Use clear instruction prompts (e.g., ‘be scientific, accurate, specific’), and re‑scope tasks iteratively.
• For literature, prefer Search/Deep research; export DOIs; verify claims against originals.
• Keep a prompt & source log (who/what/when), especially for methods and results.
• When drafts get stale, start a fresh chat and upload the latest PDF/version to avoid circular suggestions.
• Periodically prune the model’s memory of irrelevant or low‑quality items to reduce drift.

Don’t

• Paste non‑anonymised or restricted data into third‑party tools.
• Accept fabricated references or unverified quotes — always cross‑check.
• Let AI ‘decide’ methods without review; test suggested steps before adoption.
• Hide AI assistance; disclose in methods/acknowledgments per journal or LTU norms.

Workload plan (minutes add up to 80 hours required)

Submission & checkpoints

· Week 1: Lab 1 (stack + prompt log).

· Week 2: Lab 2 (SOTA brief + citation audit).

· Week 3: Lab 3 (drafted section with tracked edits).

· Week 4: Lab 4 (reproducible analysis file/notebook).

· Week 5: Lab 5 (slide outline + 2 figures/briefs + pitch).

· Week 6: Mini‑project artifact + reflective summary + demo.