Algorithms developed in this way require extensive further training to achieve desired results. Similarly, in any field, obtaining desired outcomes depends on the quantity and quality of the training data. Humans must manually classify tens of millions of records and input the results necessary for training.

Amazon developed Amazon Technologies Turk (AMT), a crowdsourcing platform capable of performing simple, repetitive tasks online, and has been using it since 2005. Google also provides a similar platform to improve search results. Recently, it is reported that tagging work was performed on 9 million YouTube videos and 14 million ImageNet photos through online crowdsourcing. It is estimated that approximately 50,000 people were involved in this effort.

Over time, the number of people researching artificial intelligence algorithms is expected to gradually decrease. This is because, unlike conventional computer software, AI can optimize itself through learning alone once it reaches a certain level.

DigiEco stated that for this reason, there is a possibility that humans could be reduced to the role of providing continuous training data to artificial intelligence. It was predicted that while online crowdsourcing expands regardless of region, gender, or age, and the number of employed people may increase, it could also lead to social conflict as it results in a number of knowledge workers being paid per task.