OPEN RESEARCH WORKSPACE / SOCIAL LAW × AI GOVERNANCE

Law, with evidence.AI, with accountability.

Exploring how social law can respond to algorithmic management, and how research on large language models can remain traceable, testable, and open to review.

Open research interests · Working tools · Reproducible methods

A SOURCE-CENTERED WORKFLOW
04

Keep the original.

Literature Library
Locate a passage02
Record a review01
Connect the next step03

From reading to an inspectable research record.

01 / BUILDING BLOCKS

Small tools. A connected workflow.

Four practical workspaces for reading, checking, tracing, and planning legal research.

01BIBLIOGRAPHY & REVIEW

01-citation-audit

Citation Audit

Inspect reference fields and duplicates, optionally compare DOI metadata with Crossref, and preserve a separate human source review.

Local checksOpt-in DOI lookup
02PASSAGES & EVIDENCE

02-evidence-atlas

Evidence Atlas

Retrieve Chinese and English paragraphs, inspect BM25 scores, and annotate how an exact source passage relates to a claim.

Source snapshots30-query benchmark
03QUESTIONS & DELIVERABLES

03-research-roadmap

Research Roadmap

Link research questions to reading, experiments, deadlines, and artifacts. Review dependencies and forecast work against real capacity.

Reading-list importResearch reflections
04THE READING WORKSPACEStart here

04-literature-library

Literature Library

Keep actual PDF, TXT, and Markdown originals alongside editable records. Browse by journal, author, year, or theme; read papers and export a source-grounded review outline.

Original files retainedComplete ZIP backupReview outline → reading plan

These tools organize and check research materials. Source authority, legal effect, and the strength of an argument remain matters for reading and human judgment.

02 / INTERACTIVE STUDIO

A little curiosity. A clearer next step.

Four free, local experiences for exploring AI collaboration, career interests, governance choices and working life. Bring a question; leave with a shareable snapshot.

YOUR AI COLLABORATION STYLE

AI Workstyle

Twenty original situations. Four collaboration axes. Explore how you prompt, check, iterate and set boundaries.

INTERESTS TO EXPLORE

Career Compass

Map six kinds of interests through original activities. Discover career families to investigate and a small experiment to try.

AI DECISIONS, HUMAN CHOICES

Algorithm Jury

Step into fictional AI dilemmas. Choose a response, examine the tradeoff, and explore the governance priorities behind your decisions.

YOUR NEXT CHAPTER

Offer Lab

Compare income, time, growth, autonomy and meaning. Change your priorities and inspect every contribution to the ranking.

Original, transparent explorations—not validated personality diagnoses, hiring predictions, or legal verdicts. No model calls or answer uploads. PNG cards are for voluntary sharing.

03 / QUESTIONS TO PURSUE

Social law meets AI governance.

Open research interests that guide the tools and a developing research agenda.

I.

Workers & algorithmic management

How do automated allocation, monitoring, and evaluation reshape workers’ ability to understand and contest decisions?

Platform labour · Procedural protection · Evidence
II.

Social protection & access

How should AI-mediated benefit services be assessed for accessibility, unequal burdens, and routes to human assistance?

Social security · Access · Human review
III.

LLM accountability

What evidence is needed to examine citation reliability, retrieval failures, and the limits of model-assisted legal work?

Source tracing · Evaluation · Reproducibility

04 / HOW THE WORK IS BUILT

Make the method inspectable.

A useful research tool should expose its sources, its rules, and its failure cases.

Read the methods note ↗
01

Traceable sources

Keep originals, paragraph positions, bibliographic candidates, and review notes connected.

02

Local-first work

Process research files in the browser. Export backups; optional DOI requests require an explicit choice.

03

Reproducible checks

Publish readable methods, automated checks, and clearly labeled synthetic examples.