Why blockchain will democratize AI (www.techradar.com)

Big tech’s dominance of AI — with Google, OpenAI, Anthropic and others controlling data, compute and talent — is creating concentration risks the industry can’t ignore: opaque data harvesting, documented model biases (a 2025 Stanford study found many LLMs perceived as left-leaning), high energy and scaling costs, and single points of failure. The article argues these problems fuel economic and societal inequality and point to real-world failures (controversial outputs from Grok and Gemini) and looming infrastructure strain (data centers could use ~20% of global electricity by 2030). Decentralized AI — combining federated learning and blockchain — is presented as a practical countermodel. Raw data stays on users’ devices while model updates or learned insights (verifiable hashes/updates) are shared and recorded on-chain, enabling auditability, distributed consensus and tokenized incentives for contributors. Workloads are horizontally distributed across participating nodes, reducing reliance on energy-hungry central data centers, lowering single-point-of-failure risk, and widening the diversity of training data to mitigate political and cultural bias. Expected use cases range from DeFi credit scoring and fraud detection to healthcare, gaming and supply chains; one market estimate forecasts growth from $550.7M (2024) to $4.33B (2034). The piece frames decentralized AI as a return of cryptography to privacy-preserving, community-driven roots with tangible technical and social implications for fairness, resilience and sustainability in AI.