Resources
A running list of external resources spanning testing and quality generally, AI included, curated as we find them.
A running list of external resources spanning testing and quality generally, AI included, curated as we find them.
40 resources
Stanford's Human-Centered AI research output, browsable and filterable by topic.
Individual model cards often document limitations, risks, and mitigations per model — worth mining for the fixes, not just the problems.
Competitions, datasets, and notebooks spanning applied ML and data science.
Trending AI research papers with code, datasets, methods, and evaluation leaderboards.
A searchable repository of real-world harms and near-misses caused by deployed AI systems.
The freshest AI research preprints as they're submitted, browsable by date.
Per-release system cards: evaluation results, safety testing, and the limitations Anthropic states itself. Read it before testing a Claude model and you start from the vendor's own list of weak spots rather than guessing at them.
OpenAI's index of system cards and post-deployment safety updates, including third-party assessments published alongside its own. The update stream matters as much as the cards — it shows what changed after release.
Model pages and technical reports for the Gemini family, covering intended use, evaluations, and stated limits per variant.
Llama's cards paired with the exact prompt formats each model expects — useful precisely because prompt-format mismatches produce failures that look like model quality problems but aren't.
Per-model capabilities, context limits, and deprecation dates. The deprecation table is the part worth watching: it tells you when a model you have tested against is going away.
The de-facto spec for what an open-model card should document. Handy as a checklist to hold a vendor card against — the gaps between the two are your test charter.
Massive collection of articles about testing fundamentals, skills in testing, test design, exploration, learning and even social sciences.
A set of patterns that helps testers to design and perform testing.
A quick read about ways to take advantage of the multiple information sources needed for ambitious testing.
By Elisabeth Hendrickson, James Lyndsay, and Dale Emery — a collection of data type attacks, web tests, heuristics and frameworks.
An evolving list of strings which have a high probability of causing issues when used as user-input data.
Become a test automation superstar — free.
All course material for BBST, the most prominent testing education.
Thematic analysis of Maaret's writings around exploratory testing.
Community curated book list for software development and testing.
The largest testing community in the world.
The largest testing community in Finland.
Context-driven testing community in Finland.
The greatest collection of international testing events.
The largest testing event in Finland.
Europe's long-running software testing conference — four days of talks and tutorials, held annually since 1993 in a different host city (Copenhagen in 2027).
Annual four-day conference in Potsdam, Germany, on testing inside agile delivery. Runs onsite and live-streamed, with the online sessions available for months afterwards.
Annual community-run conference in Tallinn, Estonia — "by testers for testers", with a practitioner programme rather than a vendor one.
A community-curated feed of testing and quality articles from across the web, contributed and tagged by Ministry of Testing members.