This is the start of a four-part series exploring how crypto and artificial intelligence (AI) intersect. Today, we tackle a big question: can crypto’s decentralized system push back against AI’s growing tendency to centralize?

For an industry based on technical fields like cryptography and game theory, crypto has an unusual knack for creating captivating stories. It’s often said that “storytellers are the most powerful people,” and crypto proves that point. If you’ve followed the space, you’ve likely seen waves of narratives rise and fall — from Bitcoin as a hedge against inflation, to decentralized finance (DeFi), NFTs in art, DAOs, the metaverse, and more recently, stablecoins as a potential product-market fit for crypto.

At its core, crypto is what’s called a general-purpose technology (GPT), much like electricity or the internet. These technologies take years to reshape industries and economies fully. While the storytelling in crypto moves fast, real-world adoption takes time and effort to tackle complex challenges like regulatory hurdles and compliance issues. The technology is ready, but now governments need to create clear rules for it to go mainstream.

The latest buzz? Crypto becoming the backbone for AI. Many believe this connection is inevitable, but we need to keep expectations realistic. The hype often overshadows the current practical applications. When we look closely at crypto and AI, the immediate overlap isn’t glamorous — but it’s incredibly important. Over the next few weeks, we’ll dive into three critical areas of their intersection:

1. How crypto can help decentralize AI to prevent excessive control by a few entities.
2. How crypto can address AI’s challenges, like verifying authenticity, improving identity systems, and restoring digital scarcity.
3. How crypto could serve as the financial infrastructure for AI systems.

Finally, we’ll spotlight one area where crypto and AI can already create significant value.

Let’s start with decentralization. AI centralization happens in three ways: controlling the computing power to train AI models, owning the data used to improve these models, and dominating the business models behind it all. Right now, AI development relies heavily on massive data centers operated by big tech companies. They have the money, hardware, and resources to scale AI quickly — leaving little room for smaller players.

The irony is that crypto itself has struggled with decentralization in some areas. Bitcoin’s original vision of “one CPU, one vote” evolved into mining being dominated by a few large hardware manufacturers. Ethereum, another major network, has also seen concentration in staking pools. If crypto couldn’t fully achieve decentralization in its own ecosystem, can it realistically decentralize AI?

Some believe that tapping into unused computing power from consumer devices — like smartphones or cars — could help. For example, parked Teslas could potentially form a distributed network for AI processing using Starlink’s connectivity. But even this idea faces challenges. Big tech companies still control the devices and have first access to this computing power.

Decentralized networks of consumer devices face technical hurdles too: ensuring reliability on untrusted hardware, distributing tasks effectively, and protecting intellectual property during training. For now, only grassroots or open-source projects might attempt such decentralized approaches — and only if they have no other choice.

Another angle is privacy. Crypto advocates argue it can give users more control over their data when interacting with AI systems. However, most people prioritize convenience over privacy, as seen in social media and other platforms. Plus, companies like Apple already offer privacy-enhancing features on their devices without needing blockchain technology. For businesses, centralized solutions by OpenAI or Anthropic offer similar security benefits without the complexity of decentralization.

When it comes to monetizing user data for AI training, the reality is underwhelming. Most user data isn’t valuable enough to warrant payment schemes. Early experiments with Web3 didn’t manage to pull users away from existing platforms, and there’s little reason to think AI-focused efforts would be any different.

Open-source AI is gaining momentum thanks to contributions from organizations like Meta and startups like MistralAI. These teams often achieve impressive results with smaller budgets by using optimization techniques and creative strategies. Open-source ecosystems powered by crypto tokens could emerge as viable alternatives in specialized AI fields — but this remains unproven.

Crypto has experimented with rewarding contributors through tokens for years, but challenges remain. Tokens often attract speculators rather than builders and can derail projects by distracting founders from building functional products.

While decentralized AI is an exciting research topic, its commercial viability is limited for now. Hybrid solutions from companies like Apple and Tesla already provide privacy benefits without decentralization’s added costs and complexity. That said, crypto shouldn’t be counted out entirely. Rapid advancements in distributed AI or a significant backlash against centralization could shift the landscape quickly.

The future might hold breakthroughs where crypto unlocks new business models for open-source AI or facilitates decentralized solutions where none exist today. And who knows? If open-source efforts lead to achieving AGI (artificial general intelligence), crypto could play a role in mobilizing resources to make it happen.

In the long run, crypto’s potential in AI may not be about decentralization alone but creating new ways to align incentives and build innovative systems. If done right, these two revolutionary technologies could pave the way for something truly transformative.