Challenges of Big Tech Dominance in AI
Big Tech companies dominate AI due to the high computational costs of building deep learning models, which smaller players cannot afford to compete with.
Models like Gemini Ultra cost around $200 million to train, forcing new entrants to rely on Big Tech for compute resources, further strengthening their position.
Big Tech’s end-to-end service offerings make it cheaper and easier to develop AI, driving up switching costs for smaller companies and consolidating their market control.
Data Monopoly and Market Consolidation
Big Tech companies hold a significant data monopoly, collecting vast amounts of data across various domains, giving them a competitive edge in AI development.
Smaller AI companies often end up selling to Big Tech, reinforcing their dominance, while public data initiatives struggle to overcome commercial capture by well-resourced actors.
The shift toward deep learning has led to Big Tech controlling most AI development, with academia playing a diminishing role in shaping research and innovation.
Need for a New Approach to AI Development
New approach to AI development is needed, one that does not replicate Big Tech’s model but challenges the existing system by focusing on theory-driven, purpose-driven, and smaller AI models
"Small AI" should be guided by domain expertise and lived experience rather than relying solely on Big Data, enabling more democratic and effective AI development.
Historically, significant advancements in various fields were driven by hypothesis testing and scientific rigor, not just big data volumes—this approach should be revived in AI.
Missed Opportunities for Change
The Global Development Compact missed an opportunity to rethink the AI paradigm, focusing on building larger data sets and providing computational power, but failing to address Big Tech monopolies or democratize AI effectively.
The current approach increases dependence on Big Tech, and unless we change course, we will remain trapped in an exploitative cycle.
COMMENTS