This post is adapted from w3w, my newsletter about emerging technology over on Substack. To receive it in your inbox every Sunday, including additional headlines from around the internet about web3, AI, and more, subscribe here. 😊
In theory, the potential intersection of web3 and AI is certainly intriguing. But in fact, real-world implementation has advanced so little that the window of opportunity for web3 to have an impact may be closing.
So says DeFi entrepreneur Jesus Rodriguez in a recent essay for CoinDesk. The co-founder and CEO of IntoTheBlock writes:
“AI innovation is compounding rapidly, and Web3 has been a passive observer. None of the major AI milestones — unsupervised pretraining, advanced fine-tuning, retrieval-augmented generation, reasoning engines, or agentic frameworks — have involved Web3 architectures in a meaningful way.
“As each new release compounds on the last, the barriers to catching up become steeper. All critical tooling, platforms, and infrastructure for building frontier models are currently centralized. Without urgent, coordinated efforts to change this trajectory, Web3-AI will be left decades behind in a field that advances in months.”
As if on cue, the headlines that broke in the days right after Rodriguez published this warning only served to underscore his point:
- Meta — unrivaled poster child for the evils of centralized tech — announced it would acquire a 49% stake in the startup Scale AI for $14.3 billion.
- Disney sued Midjourney, calling the AI image generator “a bottomless pit of plagiarism.”
Ouch.
You know, it’s almost as if the internet economy is crying out for some native way for people to immutably document identity, authorship, and provenance of intellectual property. Especially as generative AI proliferates, it would also be handy if this system were to enable easy digital payments, licensing, and collection of royalties on IP.
Or we could all just keep suing each other, running up legal bills, and see how that turns out.
Yeah. Sure. Let’s do the latter.
