Meta's new AI assistant trained on public Facebook and Instagram posts - ET Telecom

Meta has developed its new AI assistant by utilizing publicly available posts from Facebook and Instagram, while deliberately excluding private messages and sensitive personal information to respect user privacy. This approach addresses growing concerns about the ethical use of scraped data, distinguishing Meta from competitors who have faced criticism for indiscriminate data collection. By filtering out heavily private datasets and avoiding certain platforms like LinkedIn, the company aims to balance AI advancement with consumer trust and data protection standards. The development of Meta AI highlights the industry-wide challenge of managing copyrighted and private materials in large language model training. With major tech firms facing lawsuits over copyright infringement, Meta acknowledges that legal disputes regarding fair use are inevitable. However, the company maintains that their strategy of prioritizing public data and imposing strict safety restrictions on content generation, such as banning realistic images of public figures, provides a responsible framework for deploying these powerful tools. This case is relevant to open data discussions as it illustrates the complex tension between accessible public information and individual privacy rights in the age of artificial intelligence. It demonstrates how organizations can leverage open, public datasets to build sophisticated models while implementing safeguards against misuse. Understanding these methodologies is crucial for policymakers and developers navigating the intersection of open data availability, intellectual property laws, and the ethical responsibilities of AI companies.

Source: telecom.economictimes.indiatimes.com
Published on 2023-10-01