Lesson 01 setup and links

Open Teachable Machine on every laptop before people arrive, and have the GBIF map ready on the projector for the second exercise.

Before the day

• One laptop with a webcam for every three to five people.
• A browser that runs Teachable Machine. Test it on each laptop.
• A projector or a large screen for the GBIF map.
• A board or a large sheet of paper to write the GBIF numbers on.
• A plan for people who do not want their face on camera. They can train the model on hands, objects or drawings instead.

Exercise 1: Teachable Machine

• Teachable Machine image project, the page to start from.
• Teachable Machine FAQ, useful background on how training works and what happens to the pictures.

Teachable Machine can run entirely on the laptop, without webcam pictures leaving the computer, as long as nobody saves the project online. Close the exercise by deleting the project where everyone can see it.

Exercise 2: the GBIF map

• GBIF occurrence map, where records are dense and where they are missing.
• GBIF country summary for Brazil. Change the two letters in the address for other countries: US, ID, PH, KE, UG, MY.
• GBIF global data trends, charts of records over time.
• How global is the Global Biodiversity Information Facility?, an open-access paper on the gaps.

When we checked on 4 October 2026, GBIF held about 3.96 billion records. The United States had about 1.32 billion, Brazil about 45.5 million, Indonesia about 5.5 million and the Democratic Republic of the Congo about 1.6 million. Check the numbers again on the day and write them down with the date.

Questions for the closing discussion

• What did the laptop model get wrong when one person had too few pictures?
• Where does the GBIF map show the same problem across the planet?
• Who is harmed when a model is confident about well-studied places and silent about the rest?
• Who should be able to add, correct, refuse or look after the missing data?