AI to reinvent DAOs while tokenized models will become valuable: VC firm
Artificial intelligence could be the missing piece for decentralized autonomous organizations (DAOs), while trained AI models could become valuable assets on-chain, according to the co-founder of Framework Ventures.
Speaking to Cointelegraph on Sept. 5 at Korea Blockchain Week, Vance Spencer, the co-founder of the crypto-focused venture firm, shared four predictions about how AI and blockchain technology could collide.
One of the biggest impacts is for AI to finally put the “autonomous” into decentralized autonomous organizations, according to Spencer.
DAOs were founded on the concept of a decentralized collective sharing a common goal, with no overarching central authority. However, many of the biggest are still far from full decentralization or autonomy.
“It's not actually autonomous, there’s a bunch of people in the middle. It seems like AI is really the only way to actually make the DAO concept work.”In May, DAI stablecoin proprietor MakerDAO published a five-phase roadmap to upgrade its ecosystem including a strong focus on using AI to create a “governance equilibrium.”
5 phases of Endgame.
Endgame has been defined as an update that employs AI tools and open processes to enhance efficiency, resilience, and participation.
A new publication on the Maker Forum outlines its roadmap, which includes the introduction of 5 major launch phases. pic.twitter.com/8GOkIfFADi
— Maker (@MakerDAO) May 12, 2023According to MakerDAO co-founder Rune Christensen, phase three of the roadmap will launch AI tools aimed at improving and possibly automating certain governance aspects.
Christensen added these AI tools will initially help “level the playing field between deeply embedded insiders and more peripheral community members,” but eventually allow the DAO to improve its processes and decisions over time “without requiring leadership or centralized authority.
“What happens when Maker, who has a shitload of treasury, is governed by an AI?” Spencer queried.
“That AI can do really interesting things and there needs to be only limited human intervention with that,” he added.
Trained AI models could become prized
Spencer also sees a future in which trained AI models are tokenized on the blockchain.
He said an early example can be seen in the Ethereum native decentralized app and game — AI Arena — where players train an AI model to fight for them in a platform fighting game akin to Nintendo’s Super Smash Bros.
Framework invested in AI Arena's $5 million Paradigm-led seed round in 2021.
Spencer explained that in AI Arena, the players don’t control the fighters themselves but instead, the characters are controlled by AI models that are owned and trained by the player.
Here are some of the fight highlights from our Exclusive Access!
JamesRic vs Hung0 pic.twitter.com/pXJsxAJbXo
— AI ARENA | ʌı (@aiarena_) August 26, 2023He noted that while it shifts the paradigm of what a game is, the on-chain ownership of AI models is “really where this comes to life in the crypto context.”
“Probably some of the most valuable assets on-chain will be tokenized AI models, that’s my theory at least,” Spencer said.
Other use cases
Meanwhile, decentralized computing marketplaces — such as Akash Network and Render Network — could also see crypto play a part in the growth of AI.
The blockchain-based protocols work as a marketplace that allows buyers to purchase idle computing processing power from providers, which is particularly important given the current shortage of GPU chips, explained Spencer.
Related: Cathie Wood bullish on Bitcoin and AI convergence
“Actually having a network that sources and provides and bootstraps the market? Those things should work,” he said. “There are some pretty successful companies that do it that are protocols.”
Spencer also argued that blockchain technology will be important for auditing and verifying AI-provided information.
in the not too distant future, something that hasn't been validated via a blockchain will be irrelevant
"show me the data" will become "show me the proof that this data is accurate"
— Brandon Potts (@b_potts23) August 23, 2023“Say that you want to prove that ChatGPT, that specific model, is giving you an answer rather than Bard, rather than Falcon, which is UAE’s model," Spencer explained. "You can actually prove that on-chain.”
AI Eye: Apple developing pocket AI, deep fake music deal, hypnotizing GPT-4