Workspace

Our Ethical Ground

Research Remains a Human Responsibility.

AI can help us search, analyze, and write. The work still needs permitted inputs, evidence we can inspect, and people who can explain and correct its claims.

Our Principles

The Distinction

Assistance is Not Evidence.

A suggested paper is a lead until someone verifies and reads it. Generated code is a draft until someone tests it. An AI summary can guide attention, but the source must support the claim.

The more a task can change a finding or affect a person, the stronger its oversight should be. Research should remain understandable, checkable, and correctable even when some steps are delegated.

How We Work

Five Principles for AI Assisted Research

01

People Set the Purpose

Researchers decide which questions matter, which methods fit, and what conclusions the evidence supports. AI may suggest possibilities; it does not make those judgments for them.

02

Permission Comes First

Before using a tool with participant data, confidential work, or licensed material, check consent, ethics approval, agreements, and the tool’s data handling. Leave material out when its use is not authorized.

03

Keep the Work Traceable

Preserve the sources, inputs, code, decisions, and outputs needed to understand how a result was made. Record meaningful AI contributions and disclose them under the rules of the course, institution, funder, or journal.

04

Check Claims against Evidence

Verify references in the original source, inspect summaries against the text, and test generated code and analysis against data and assumptions. A fluent answer or successful run is not validation.

05

Accountability Stays Human

A person or responsible team must be able to explain, defend, and correct the work. AI cannot provide consent, approve a study, serve as an author, or take responsibility for a published claim.

Our Test

Can we show what AI did, why it was allowed, how we checked it, and who stands behind the result?

In practice

The Check Follows the Task.

AI Can AssistResearcher Checks

Literature

Suggest search terms and summarize candidates

Read the papers behind central claims; verify citations and missing perspectives.

Data

Help transcribe, clean, or classify permitted material

Confirm consent and access; compare outputs with originals and preserve raw inputs.

Analysis

Draft scripts, diagnostics, and figures

Run and test code; review transformations, assumptions, uncertainty, and figure provenance.

Writing

Outline, revise, and clarify prose

Check claims, quotations, and numbers; disclose substantial assistance where required.

Beeblio’s role

A workspace for work you can inspect.

Beeblio brings sources, project files, analysis, and writing together so researchers can inspect the materials behind AI assisted work. That makes review possible; it does not perform the ethical or scholarly judgment for you. Check your study’s approvals and the rules that apply to your data before bringing material into any AI workflow.