The mission of the Mozilla Foundation is to make the Internet a global public resource, open and accessible to all. In 2017, she launched an open source initiative: The State of the Internet Report, which aims to document and analyze the state of the Internet based on various studies from multiple sources. This year’s report focuses on one important topic for the Internet and humanity: artificial intelligence.
AI is part of our daily lives, but just like the Internet, GAFAM and some of the rich and powerful are tightening their grip on AI to increase their profits.
The Foundation is reminded of the dangers of discrimination and surveillance that automation brings, and of over-reliance on it for decision making with huge implications, and that proponents argue there is a better way to create, deploy and understand the potential of AI.
“Who has power over AI? ”, “Who transfers this power? are the central questions of this report, which over the course of several months, more than 150 people worked on. They built on the extensive work of Mozilla Foundation staff, fellows, and grantees to find more robust AI in products, policies, and coding.
Along with the 2022 report, Mozilla explores a wide range of topics, including surveillance, labor, health, geospatial data or social media disinformation, in five biweekly podcasts featuring 19 people from a dozen different countries.
Who has power over AI?
Companies across industries are turning to AI to collect and analyze data to tailor their offerings. Those with enough resources to invest in AI have a competitive advantage. Meanwhile, countries with access to large amounts of data and high computing power are strengthening their dominance in software and hardware, influencing how AI is deployed around the world.
In terms of private investment in AI, the United States dominates. Next come China, Britain, Israel, Germany, Canada, France and then India.
Use different datasets
The cost of training machine learning systems has gone down and data availability has gone up. However, the authors of the report were able to see that the same datasets from just a few countries are most often used to evaluate the performance of machine learning models around the world.
As such, more than half of the datasets used to benchmark AI performance across over 26,000 research papers came from just 12 established institutions and technology companies in the US, Germany, and Hong Kong, China.
The report notes that a large, frequently reused dataset does not guarantee better machine learning than a smaller, purpose-built dataset. On the contrary, it often reflects the racist, sexist stereotypes of the Internet content with which it has been trained.
The companies that will benefit the most from AI are arguably the biggest tech companies in the world. Their revenues have skyrocketed during the COVID-19 pandemic, especially for Microsoft in 2021, and some of them are now among the wealthiest companies in the world.
They influence Internet users by recommending publications, purchasing products, etc.
According to the Digital Rights Rankings, “There is little to no transparency about how companies test and deploy algorithmic systems and use our personal data, including with respect to curation, recommendations, and rankings.”
Recommender systems determine which social media channels people will have access to and what content will be rewarded with ad revenue. In a YouTube crowdsourcing study by Mozilla, 71% of the videos that people said they “regret” watching were recommended to them by an algorithm.
Advertising and social media can be used to spread propaganda and misinformation. In the absence of transparency and cooperation with researchers, a global industry of companies and organizations engaged in the exchange of secret messages thrives.
How to establish trust?
New methods and techniques for maintaining privacy, recording the origin of data, enforcing AI ethics, auditing algorithms, and user empowerment are some of the many traps being considered.
This requires cooperation. With this report, and the accompanying podcasts and stories, Mozilla is calling on technology developers and policy makers to start a conversation and take action.
2022 Internet Health Report https://2022.internethealthreport.org.