Literature Research and Analysis

The History of Social Security in Switzerland

A guided exercise on systematic literature research and problem-oriented literature analysis, with a critically controlled use of generative AI.

Author

Affiliation

Moritz Mähr

University of Bern

WarningAutomated Translation

This exercise was translated automatically from the German original and can contain errors. Consult the original version in case of doubt.

Overview and Didactic Goal

This exercise introduces systematic literature research and problem-oriented literature analysis. The goal is that you can reconstruct the state of research on a historical topic. One focus lies on the critically controlled use of generative AI as a heuristic tool, which widens the search space and structures texts.

Case study: the history of social security in Switzerland

The exercise follows a circular process of search and analysis. It starts from a broad field (social security). From it you develop step by step a clear problem focus, a precise research question and a reproducible corpus of relevant secondary literature. Manual phases of search and analysis alternate with AI explorations, and you validate each exploration critically.

At a Glance

Item	Value
Duration	about 1 working day (6 to 8 hours)
Level	Intermediate
Core path	1 → 2 → 3 → 5 → 6 → 8 → 9
Optional deepening	4 (special paths), 7 (mapping the debate), the advanced track
Hand-in (minimum)	Search log: 12 entries. Reject log: 5 entries. AI log: 4 entries.

Split the working day over two or three sessions. The core path leads you to a precise research question. The optional deepening costs about 2 hours more.

CautionPlan time for the access to the literature

Order the titles that you lack early. An interlibrary loan often takes several days. The duration above does not include this waiting time.

What You Need

Subject knowledge

• A basic understanding of historical research methods
• Basic knowledge of work with generative AI, above all prompting

NotePrompt Engineering

If prompting is new to you, we recommend that you work through the exercise Prompt Engineering first.

Capabilities of the AI system

The system must be able to do the three following things:

• Search the web and support each finding with an address. You need this in step 2 and in step 4.
• Read a file that you upload, for example a list of hits or an article.
• Deliver an answer in a format that you prescribe, for example as a table.

Other tools and access

• A reference manager. You export your list of titles from it.
• An account for the library catalogue of your university. You need it for step 3 and for the interlibrary loan.
• One file for your search log and one file for your AI log.

NoteAccount, cost, data protection and law: settle this before you start (expand)

You can do every exercise with AI systems from different makers. What a system can do depends on the product and on your account. Settle the five following points before you start.

Cost

A free account is enough for short text tasks. For some capabilities you need a paid account. This mostly concerns the search on the web, the reading of long files and the built-in reasoning mode. Check first whether your university provides access.

Tools

Check which tools your account unlocks. Can the system search the web? Can it read a file that you upload? Can it call external services? Each exercise names the capabilities that it requires.

The section Check and classify capabilities explains five newer capabilities. It also describes what each capability means for your work with sources.

Memory

Some systems carry over statements from earlier sessions or from a project context. Check the settings for memory, projects and chat history. A new session therefore guarantees no empty context.

Account: institutional or private

Check which account you work with. The rules differ in three points.

Point	Account of your institution	Private account
Contract	Your university concludes it and shares the liability.	You conclude it yourself and carry the liability alone.
Training	A contract often excludes the use of your inputs.	Many providers use inputs for training.
Permitted material	The university rules which data you can enter.	You have no such rule and carry the risk yourself.

Ask the IT office or the library of your university when you cannot find the rules. Record the answer in your AI log.

CautionCheck every piece of material before you upload it

Do not upload material that you have no right to pass on. An upload leaves your computer and you cannot take it back.

These materials stay outside the AI system:

• Personal data of living persons, for example from interviews, from private papers or from holdings under a protection period.
• Digitized items whose terms of use forbid a disclosure.
• Unpublished texts of other persons, for example manuscripts or reviews.
• Licensed full texts, when the license of your library forbids a disclosure.

In case of doubt, work with a short excerpt or with an anonymized version. You can also use a local model, because the material then stays on your computer.

Law: data protection and disclosure

Two areas of law concern your work directly. This section gives an orientation and no legal advice.

Data protection. In the EU the General Data Protection Regulation applies. In Switzerland the revised Data Protection Act applies (Verordnung (EU) 2016/679 Zum Schutz Natürlicher Personen Bei Der Verarbeitung Personenbezogener Daten (Datenschutz-Grundverordnung) 2016, Bundesgesetz Über Den Datenschutz (DSG) 2020). Both laws protect personal data of living persons. The text of the law itself is binding. Article 6 GDPR requires a legal basis for every processing operation, and an upload is a processing operation.

The role of the provider decides the classification. A provider can act as a processor and process your data only on your behalf. Article 4 point 10 GDPR expressly excludes a processor from the third parties. This role needs a contract under Article 28 GDPR. Without such a contract, a provider uses your inputs for its own purposes. An upload is therefore not always a disclosure to a third party, but without a contract you must assume that it is. Check the contract that applies to your account.

Research is subject to safeguards, but it gets no exemption. Article 89(1) GDPR requires appropriate safeguards, for example data minimisation and pseudonymisation. Article 89(2) GDPR permits the legislator to derogate from single rights of the data subject. The information of the data subjects can also fall away, namely under Article 14(5)(b) GDPR. That applies only when the information proves impossible or would involve a disproportionate effort. Research therefore does not release you from the information duty in general. Settle your individual case with the data protection office of your university.

The European Data Protection Board presented a guideline on research in 2026 (European Data Protection Board 2026). This guideline was in public consultation until June 2026. It is therefore not applicable law, but an aid to interpretation. For deceased persons, the protection periods and the rules of use of the archive apply.

Disclosure. The AI Act of the EU governs the transparency of generated content (Verordnung (EU) 2024/1689 Zur Festlegung Harmonisierter Vorschriften Für Künstliche Intelligenz (Verordnung Über Künstliche Intelligenz) 2024). Article 50 applies from 2 August 2026. It separates two addressees: the provider of an AI system and the deployer. You are a deployer as soon as you use an AI system professionally. For purely private use you are not a deployer, under Article 3 point 4.

As a deployer you carry two separate duties from Article 50(4).

1. Deep fake. If an AI system generates or manipulates an image, an audio recording or a video, and the content is a deep fake, then you disclose that. Under Article 3 point 60, a deep fake resembles existing persons, places or events and falsely appears authentic. A generated image without this resemblance to reality does not fall under this duty.
2. Text on public matters. If you publish generated text to inform the public on a matter of public interest, then you disclose that. The duty falls away when a person reviewed the text and holds the editorial responsibility for the publication.

Article 50(4) therefore requires no disclosure for every generated image and for every generated text. Article 50(2) addresses the provider instead. The provider must mark generated content in a machine-readable format. This duty does not lie with you, and it does not replace your own disclosure.

ImportantThe legal disclosure does not replace the academic one

The legal disclosure and the academic disclosure are two separate duties. Your university requires a declaration even when the law requires no disclosure. The exercise Citing shows how you declare your AI use.

Record your answers to these five points. They belong to the description of your work environment.

Learning Objectives

After this exercise you can:

• give reasons for the function of literature research and literature analysis in historical research,
• translate a topic on Swiss social security into a search strategy,
• search systematically (catalogues and databases) and unsystematically (snowballing), and document both,
• use AI as a heuristic aid for the search (expansion of keywords, design of queries, screening),
• judge the relevance, the argument, the theory and the methods of secondary literature in a structured way,
• identify the central lines of debate and the gaps in research, and derive a precise research question from them,
• run your reference management reproducibly and with correct citations (reference manager, BibTeX, search log, AI log).

What Are Literature Research and Literature Analysis?

Literature research is the systematic identification of the relevant secondary literature, and of the relevant tertiary resources, to map the state of research.

Literature analysis is the problem-oriented, critical reading of this literature. It reconstructs arguments, references to theory, methods, bases of evidence and controversies, instead of extracting “facts”.

Case study: the history of social security in Switzerland (the welfare state, the branches of social insurance, federalism, associations, the gender order, transnational transfers).

NoteNote on the search online

Digitization advances, but historical research is not possible online alone. (Milligan 2019) Secondary literature that is out of print, and primary sources that no one has catalogued yet, often demand a physical search in libraries and archives. Plan your access (library, interlibrary loan, archive). Note that platforms and catalogues structure visibility through their logic of cataloguing, and that they guarantee no completeness. (Putnam 2016)

CautionWhen your system searches on its own

Some accounts offer deep research. The system then searches the web on its own, reads pages and delivers a report with references. You can use this report as a starting point, but not as the state of research.

Three reasons speak for a complete counter-check:

• Such a workflow plans and acts over several steps. An error from an early step travels into all later steps, and the performance reports about such systems are often not reproducible (Kapoor et al. 2024).
• Literature references from AI systems hold fabricated and faulty titles to a considerable degree (Walters and Wilder 2023).
• The report does not name its search strategy in full. You therefore cannot judge its coverage.

Check every title in the catalogue of your library. Record the search in the search log and the AI use in the AI log. The section Check and classify capabilities explains these workflows.

TipSearch your own documents faster

Some accounts permit retrieval from your own documents. You deposit your PDF files, and a search step selects the fitting passages from them (Lewis et al. 2020). That helps in the literature analysis in step 6, because you find the fitting passages in your own material faster.

The retrieval does not bind the answer to your collection. It only widens the context by the passages that the search found. The system can still mix in knowledge from the training, and the search step does not find every relevant passage. A location reference also supports only the location, and not the statement. Open therefore every page that the system names and read it yourself.

A deposited collection is an upload. Check therefore the license of your library before you deposit licensed full texts. The section “What you need” names the rules.

Structure of the Exercise

Each step holds:

• an objective,
• concrete tasks, including optional AI prompts,
• a work assignment and a reflection assignment.

1. Formulate the problem and the search horizon
2. Orientation knowledge from tertiary resources (a frame of debates and terms)
3. Systematic search I: catalogues (coverage and the logic of retrieval)
4. Systematic search II: journals, media and special paths (context, reviews, terms)
5. Snowballing: cite backwards and forwards (control the canons and the blind spots)
6. Screening and quality assessment: relevance, argument, method, evidence
7. Map the historiographic debate and write the state of research (condensed, sensitive to evidence)
8. Gap in research and delimitation: from the state of research to the research question
9. Reproducibility and records

1. Formulate the Problem and the Search Horizon

Goal

Move from the broad field (social security) to a clear problem focus and to a first hypothesis for the search.

Task (Without AI)

Formulate in 5 to 8 sentences:

• the period of study (for example 1880–1950, 1938–1948 or 1970–2005)
• the branch (for example old-age provision, health insurance, disability, unemployment)
• the assumed logic of conflict or mechanism (for example federalism and the referendum, associations, expertise, gender, transfer)
• the expected fields (history, political science, legal history, social policy)

Task (AI Exploration)

I work on the topic "the history of social security in Switzerland".
Suggest 6 problem-oriented variants of a focus that a historian can work on (each with a period, a focus on actors or arenas, and a mechanism).
Give 8 search terms per variant (DE/FR/EN, historically plausible terms) and 3 expected literature types (monograph, article, chapter in an edited volume).
No invented titles.

Critical Check

Mark in the answer:

• the anachronistic categories (mark them as search anchors in hindsight),
• the implicit assumptions of theory (for example “welfare regime”),
• the missing perspectives (for example gender, the cantonal level, international organizations and transfer).

2. Orientation Knowledge from Tertiary Resources (A Frame of Debates and Terms)

Goal

Build a robust frame of context (chronology, terms, institutions) as the basis for a targeted literature search.

Core Resources

• Platform:
  • History of Social Security (Switzerland) — Home
  • History of Social Security — Synthesis
  • History of Social Security — Bibliography
• Entry points in the Historical Dictionary of Switzerland:
  • Historical Dictionary of Switzerland: Social Insurance
  • Historical Dictionary of Switzerland: Social Policy
  • Historical Dictionary of Switzerland: Old Age and Survivors’ Insurance (AHV)
  • Historical Dictionary of Switzerland: Social Security

Task (Without AI)

Read, in skimming mode:

• one stage of the synthesis on the platform,
• one article of the dictionary, or several ones briefly.

Extract:

• 8 to 12 key terms, including historical terms,
• 6 “fixed points” (events, laws, institutions),
• 5 “markers of debate” (for example federalism as a veto point, the relation between state and private, gender, transfer, expert knowledge).

Task (AI Exploration)

Start from the following terms and fixed points: <insert your 6 fixed points and 5 markers of debate here>
1) Build a map of the debate (4–6 axes) on the history of social security in Switzerland.
2) Assign 5–8 types of author and work to each axis that one would typically expect (without inventing concrete titles).
3) Derive 10 search strings for catalogues from it (Boolean, phrases).

Critical Check

Validate the “axes of debate” with at least 2 independent entry sources (the dictionary, the platform, the tables of contents of edited volumes).

3. Systematic Search I: Catalogues (Coverage and the Logic of Retrieval)

Goal

A reproducible, documented search in large catalogues, and the build-up of a starting corpus.

Primary Search Location

• swisscovery (SLSP)

Task (Without AI)

Run 6 searches in swisscovery:

• 2 with broad terms,
• 2 with precise phrases and filters,
• 2 with several languages (DE/FR/EN) and variants of names.

Document for each run:

• the query (exact), the filters, the number of hits, the export date,
• 3 criteria of selection (relevance, period, publication type),
• 3 criteria of exclusion.

Task (AI Exploration)

Run the searches heuristically with AI. Give enough context.

You are a historical research assistant with knowledge of Swiss social, economic
and political history.

Context:
- Research topic: the history of social security in Switzerland
- Period: [for example 1880–1950]
- Goal: build a first and broad starting corpus of secondary literature

Task: search for relevant secondary literature.

Conditions:
- Invent no concrete titles or authors.
- Pretend no knowledge of a catalogue.
- Mark uncertainties explicitly.

Critical Check

• Adopt no title from an AI suggestion before you find it in the catalogue.
• For each title that you adopt, note in one sentence why you adopted it.

4. Systematic Search II: Journals, Media and Special Paths (Context, Reviews, Terms)

Goal

Make the debates and the state of research visible through articles, reviews and periodicals.

Search Locations (Examples)

• Journal platforms: E-Periodica — Swiss Journals Online
• Newspapers (for paths of discourse and reception): e-newspaperarchives.ch — Digital Newspaper Archives

Task (Without AI)

Choose 2 journal titles or corpora of periodicals and search for:

• “Sozialversicherung”, “AHV”, “Invalidenversicherung”, “Krankenversicherung”,
• plus your term of focus.

Extract:

• 5 relevant articles or reviews,
• the terms and frames that appear again and again (for example “Lasten”, “Solidarität”, “Missbrauch”, “Drei-Säulen”).

Task (AI Exploration)

From these 5 abstracts and titles: <insert your 5 abstracts or titles here>
1) Cluster them by the logic of the question (institutional, centred on actors, history of ideas, political economy, gender).
2) Suggest 2 follow-up keywords per cluster (DE/FR/EN) and name 1 typical "blind spot" for each.

Deliverable: a mini cluster of 5 items and an updated list of keywords.

5. Snowballing: Cite Backwards and Forwards (Control the Canons and the Blind Spots)

Goal

Go into depth from a small core corpus through footnotes and bibliographies. Make the bias of the canon visible at the same time.

Task (Without AI)

Choose 3 “seed” texts, of types as different as possible: an overview, a monograph and an article.

For each seed:

• 10 relevant references from the bibliography or the footnotes (backwards),
• 5 newer works through a citation search (forwards, for example in a scholarly search engine or on publisher pages).

Document:

• why you adopted each reference (1 sentence),
• whether you widened or narrowed an axis of the debate.

Task (AI Exploration)

Here are 10 references from the bibliography of a seed text: <insert your 10 references here>
Order them by:
(a) the axis of debate, (b) the method, (c) the expected base of sources.
Mark which 3 would be indispensable for a section on the state of research, and give reasons.

Critical Check

Keep a section reject log in your search log: at least 5 hits that you rejected although they looked plausible, with the reason for the rejection (wrong period, wrong arena, popular science only, and so on).

Deliverable: a snowball list of at least 30 references and a section reject log with at least 5 entries.

6. Screening and Quality Assessment: Relevance, Argument, Method, Evidence

Goal

Do not only collect literature, but select it analytically.

Task (Without AI)

Build a rubric of assessment on one page, for example:

• the research question or thesis
• the theory and the terms (explicit or implicit)
• the method (based on archives, discourse analysis, comparative, quantitative)
• the base of sources (which primary sources? which gaps?)
• the contribution (what is new? what does the text dispute?)
• the fit with your focus

Apply it to 6 titles from your corpus, with a short excerpt of 150 to 250 words per title.

Task (AI Exploration)

Use this rubric: <insert your rubric here>
Build a structured short annotation (max. 180 words) for each of the following titles and abstracts.
Include (1) the assumed contribution, (2) the methodological approach, (3) a possible bias or limitation.
If information is missing, mark it expressly as "unclear" and do not add anything.

Critical Check

Take a sample: for 2 titles, check against the introduction and the conclusion whether the AI structure and your own reading agree.

Deliverable: 6 annotated entries and the rubric.

7. Map the Historiographic Debate and Write the State of Research (Condensed, Sensitive to Evidence)

Goal

Turn the corpus into a reconstructed “conversation”: who argues how, and why?

Task (Without AI)

Write a memo on the debate of 400 to 600 words:

• 2 to 3 main controversies,
• the central positions,
• the typical disputes over evidence or method,
• the open flanks (untouched, underexposed, contested).

Task (AI Exploration)

Here is my memo on the debate: <insert your memo here>
1) Identify the implicit assumptions (agency against structure, state against private, normativity, diagnoses of the present).
2) Suggest 2 alternative ways to structure it (for example by arenas instead of by topics).
3) Name 5 targeted questions that I should put to the literature, so that I can formulate the gaps in research precisely.

Deliverable: the memo on the debate and the 5 questions.

8. Gap in Research and Delimitation: From the State of Research to the Research Question

Goal

Derive a research question that you can work on and that is analytical from the state of the literature, and not from your interest alone.

Task (Without AI)

Formulate:

• 1 statement of the gap in research (at most 3 sentences): “Research has shown X; Y stays unclear, because …”
• 2 research questions of one sentence each that are plausible on the basis of the sources and the literature.

Task (AI Exploration)

Start from this gap in research: <insert your statement of the gap here>
Formulate 4 variants of an analytical historical research question.
For each variant:
- a possible delimitation of the case (time, space, arena),
- the expected contribution (which debate does it address?),
- a field of risk (for example access to sources, anachronism, causality that is too broad).

Deliverable: 1 gap in research, 1 research question with priority, and 1 section on risk of 150 to 200 words.

9. Reproducibility and Records

Goal

Work with literature as a traceable research process, oriented on the FAIR principles and safe for citation.

Minimum Requirements

• Search log, section on the search: date, system, query, filters, hits, selection rules (12 entries or more)
• Export from the reference manager or in BibTeX: a versioned state with a date, and a strategy for duplicates
• Search log, section on decisions: why you adopted or rejected a title (in keywords)
• AI log (appendix): 4 entries or more, each with the prompt, the shortened answer, your corrections or rejections, and the purpose of use

WarningAI as a tool, not as a source

In this exercise, AI can help you to widen the search space and to structure texts. It does not replace:

• the bibliographic verification (catalogue, DOI),
• the check of the base of evidence with source criticism,
• your historiographic judgement.

Learning Outcome

At the end you hold these artifacts:

1. A statement of focus and a map of the debate (1 to 2 pages)
2. A search log with a section on the search (12 entries or more) and a section reject log (5 entries or more)
3. A short annotated bibliography (10 to 15 titles)
4. A memo on the debate (400 to 600 words)
5. The gap in research, the research question and the field of risk (about 1 page)
6. An AI log (appendix, 4 entries or more)

One methodological insight comes on top: you can give reasons for which hits a catalogue shows, and which ones it hides systematically.

Optional: Advanced Track (Digital Methods)

• Export the metadata (title, author, year, keywords) from the catalogue or from your reference manager.

• Build a simple map of the bibliography:

  • the distribution over time (years of publication),
  • the languages and publication types,
  • a network of co-citations or of subject headings (with care: subject headings in a catalogue are curated and carry bias).

• Document the workflow and the data (a README file and notes on the license and on reuse).

Further Resources

• Ranke.2: Lessons Overview – Learning modules on the web, social media and digital change.
• Knowing Machines: Models All the Way Down – Critique of training data and its parallels to source criticism.

Bibliography

Bundesgesetz Über Den Datenschutz (DSG) (2020). https://www.fedlex.admin.ch/eli/cc/2022/491/de.

European Data Protection Board. 2026. Guidelines 1/2026 on Processing of Personal Data for Scientific Research Purposes. Version 1.0. European Data Protection Board. https://www.edpb.europa.eu/our-work-tools/documents/public-consultations/2026/guidelines-12026-processing-personal-data_en.

Kapoor, Sayash, Benedikt Stroebl, Zachary S. Siegel, Nitya Nadgir, and Arvind Narayanan. 2024. “AI Agents That Matter.” July 1. https://doi.org/10.48550/arXiv.2407.01502.

Lewis, Patrick, Ethan Perez, Aleksandra Piktus, et al. 2020. “Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks.” May 22. https://doi.org/10.48550/arXiv.2005.11401.

Milligan, Ian. 2019. History in the Age of Abundance? How the Web Is Transforming Historical Research. McGill-Queen’s University Press. https://doi.org/10.2307/j.ctvggx2kh.

Putnam, Lara. 2016. “The Transnational and the Text-Searchable: Digitized Sources and the Shadows They Cast.” The American Historical Review 121 (2): 377–402. https://doi.org/10.1093/ahr/121.2.377.

Verordnung (EU) 2016/679 Zum Schutz Natürlicher Personen Bei Der Verarbeitung Personenbezogener Daten (Datenschutz-Grundverordnung), Amtsblatt der Europäischen Union 1 (2016). https://eur-lex.europa.eu/eli/reg/2016/679/oj/deu.

Verordnung (EU) 2024/1689 Zur Festlegung Harmonisierter Vorschriften Für Künstliche Intelligenz (Verordnung Über Künstliche Intelligenz), Amtsblatt der Europäischen Union (2024). https://eur-lex.europa.eu/eli/reg/2024/1689/oj/deu.

Walters, William H., and Esther Isabelle Wilder. 2023. “Fabrication and Errors in the Bibliographic Citations Generated by ChatGPT.” Scientific Reports 13 (1): 14045. https://doi.org/10.1038/s41598-023-41032-5.