Meta will train artificial intelligence by monitoring every move of employees

Meta is turning to a remarkable method in its search for new data sources to develop artificial intelligence models. The company plans to record the keyboard keys and mouse movements of its employees regarding computer use and use this data to train artificial intelligence systems. This development, first brought to the agenda by Reuters, reveals how technology companies are expanding their data collection methods to create more competent models.

In the statement made by the company to TechCrunch, it is stated that such data is needed so that the developed artificial intelligence systems can learn real user behavior. In the statement, it is stated that they want to develop artificial intelligence agents that can understand daily transactions performed on the computer, so interactions such as mouse movements, clicks and menu navigations are examples. In this context, it is reported that an internal tool running on certain applications will be activated and this tool will record user interactions. However, it is emphasized that there are various security measures to protect sensitive content and that the collected data will not be used for different purposes.

Meta’s approach aims to increase the diversity of data needed for artificial intelligence systems to work more efficiently. In addition, it is evaluated that models that can directly observe users’ behavior on the computer can provide more effective results, especially in the field of automation and digital assistants. However, this method also brings employee privacy and data security issues back into discussion.

The search for data for artificial intelligence training is expanding

Recently, similar examples have begun to be seen more frequently in the technology sector. In another news that appeared last week, it was stated that internal communication records of old startups, such as Slack correspondence and Jira task records, were evaluated for artificial intelligence training. It is stated that such content provides rich data in terms of natural language processing models. Despite this, questions such as within what limits the data in question is used and to what extent it is anonymized remain on the agenda.

In addition, the increasing need for data in the artificial intelligence development process directs companies to alternative sources. Considering corporate internal data as a potential source in addition to open internet data expands the scope of data collection practices. However, there is no clear framework on how this approach will be regulated in the long term and how user rights will be protected.

This application planned by Meta can enable artificial intelligence models to produce results closer to real usage scenarios. Despite this, this use of employee data stands out as a factor that directly affects internal transparency and trust issues. Therefore, if similar practices become widespread, it is expected that both legal regulations and company policies will become more clear.