Concerns about Big Tech’s dominance in AI

Source: The post concerns about Big Tech’s dominance in AI has been created, based on the article “Democratising AI needs a radically different approach” published in “The Hindu” on 23rd November 2024

UPSC Syllabus Topic: GS Paper 3-Science and Technology – Developments and their applications and effects in everyday life.

Context: The article highlights concerns about Big Tech’s dominance in AI due to high costs, data monopolies, and integrated services. It suggests shifting focus from “bigger is better” AI models to smaller, purpose-driven AI guided by theory, expertise, and progressive goals.

For detailed information on Regulating Big Techs In India read this article here

What is the concern about Big Tech’s dominance in AI?

1. High Computational Costs: Building deep learning models is expensive. For instance, the Gemini Ultra model cost $200 million to train, making it nearly impossible for smaller companies to compete.
2. Dependence on Big Tech: Smaller players rely on Big Tech for compute credits, deepening their dominance.
3. End-to-End Services: Big Tech offers integrated tools for tasks like data preparation and algorithm access, making their platforms more convenient and cost-effective.
4. Data Monopoly: Big Tech collects vast, continuous data streams across domains and geographies, creating a competitive edge. Open data initiatives often fail as Big Tech leverages public data more effectively.
5. Control Over Research: Big Tech now leads academic research, surpassing universities in publications and citations, influencing AI’s direction.
6. Missed Opportunities: Initiatives like the Global Development Compact fail to rethink the paradigm, reinforcing the “bigger is better” model without addressing systemic issues.

What is the Proposed Shift in AI Development?

1. Emphasis on Small AI: The shift suggests moving away from massive data-driven models, like Gemini Ultra, which costs about $200 million to train, towards smaller, targeted models that focus on specific societal needs.
2. Theory of Change: This approach advocates for AI development guided by causal relationships and hypotheses testing, rather than sheer data volume.
3. Domain Expertise: It prioritizes the use of domain knowledge and lived experiences to inform AI development, ensuring that models are relevant and effectively address real-world challenges.
4. Purpose-Driven Models: By focusing on specific goals, these smaller models can be more sustainable and democratic, offering alternatives to Big Tech’s monopolistic practices.

Question for practice:

Examine the concerns associated with Big Tech’s dominance in AI and the proposed shift towards smaller, purpose-driven AI models.