Starter Kit. Idea 304 of 500. Cluster 15
Data labelling and AI-training services, which is the new garment work and pays better
The mechanism, from the book
The proof exists in Malabe. What is missing is a thousand small versions of it.
The first step
Complete a paid test task on a platform that pays per labelled item, and record your hourly earnings and accuracy score. Then gather five people you trust, train them on the same task, and offer the team's capacity and quality record directly to a small AI company abroad that posts labelling work. A direct contract above the platform rate is the proof.
- Who pays first
- A small AI company abroad that needs labelled data checked to a known accuracy and pays more for a steady team than for anonymous piecework.
- What leaves today
- Labelling work is done here as piecework through foreign platforms, which keep the margin, instead of by teams that sell direct.
Ask first
- Data Protection Authority. Ask what rules apply when your team handles personal data inside labelling work.
- Information and Communication Technology Agency. Ask whether any training or support exists for data and AI service firms.
Check before you spend
What the client's contract says about confidentiality, whether your team works as employees or contractors, and how data is protected on the machines used.
Find out these three numbers
- What does a platform pay per hour at your team's accuracy?
- What do small AI companies pay for labelling under direct contract?
- How many labelled items can one trained person finish an hour?
A comparison from the book
Costa Rica. Chapter Eleven
"Costa Rica lost the tenant and kept the capability"
Train the team so the skill stays here even if one platform or one client leaves.
The test
Does it keep value that now leaves the island, or build the proof that lets somebody else do so?
From the appendix of Why Not Sri Lanka? by Dr Maheshika Halbeisen. The idea and the mechanism are the book's; this kit was written for the site.