Crypto-based AI markets, and AI financial analysis – Cointelegraph Magazine

This week we present real use cases for AI and crypto every day. Get two for the price of one today: Blockchain-based AI marketplaces and financial analytics.

Mixing AI and crypto might not sound like the most exciting use case, but both recent co-founder Illia Polosukhin and Framework Ventures founder Vance Spencer cite blockchain-based marketplaces and computing data as their top picks for AI.

AI is an incredibly fast-growing industry that requires computing power. Microsoft alone is said to be investing $50 billion in data center infrastructure by 2024 to handle demand. AI also requires vast amounts of raw data and training data labeled by human categories.

Polosukin believes that decentralized blockchain-based marketplaces are an ideal solution to help decongest the required hardware and data.

“You can use it [blockchain] They are more equal to building more efficient marketplaces,” he told the magazine, explaining that AI projects currently have to negotiate with one or two large cloud providers, such as Amazon Web Services. Still, it's difficult to get the necessary capabilities due to the lack of Nvidia's A100 graphics processing units.

Spencer cites blockchain-based marketplaces as the current number one use for AI resources.

“The first is bringing real GPU chips,” he said. “With a huge shortage of GPU chips, how do you produce them? [without] Do we have a network that generates and delivers and hosts the market?

Spencer highlights the Akash Network, which provides a marketplace for decentralized computing resources across the cosmos, and the Render Network, which provides a distributed GPU approach.

“At this point, there are some pretty successful companies that are actually working on protocols.”

Another example of a decentralized marketplace that offers cloud computing for AI is Aleph.im. Token holders in the project have access to computing and storage resources to run projects.

Libertai.io, a decentralized large-scale language model (LLM) running on Aleph. While you might think that decentralization will slow AI down to the point where it can't take off, Alephim founder Moshe Malavach says that's not the case.

“Here's the thing: For one user, the whole inference (using a model to generate data) runs on one computer. It is decentralized by entering random computers on the network. But then, time for your question is centralized. So it can be fast.”

Another blockchain-powered AI marketplace is SingularityNET, which offers a variety of AI services — from image generation to coloring old pictures — that users plug into models or websites.

What Spencer is most excited about is blockchain-based AI marketplace tokenization and trading of AI models. Structure has invested in AI Arena, a Super Smash Brothers-style fighting game where users train AI models to fight each other. The models are stored as invulnerable tokens and can be bought, sold or rented. “I think it's great,” he said. “Native monetization of crypto is exciting, but so is ownership of these models.”

“I think one day maybe some of the most valuable models — some of the most valuable assets on the chain — will be marked by AI models. That's my theory, at least.”

Don't believe the hype: you can now create components, data, and compute with traditional Web2 marketplaces.

Bonus Use Case: Financial Analysis

Anyone who has attempted to decipher the ocean of information produced by chain financial transactions knows that while it is one thing to keep consistent and transparent records, it is another to be able to analyze and understand them.

AI analytics tools are perfectly suited to summarize and interpret patterns, trends and anomalies in the data, and can suggest strategies and insights to market participants.

For example, Mastercard's Safetrace Armada platform recently partnered with AI company Feedzai to use the technology to analyze, identify and block fraud or money laundering-related transactions across 6,000 exchanges.

Elsewhere, the GNY.io machine learning tool tries to predict the volatility of the top 12 cryptocurrencies and the range report uses ChatGPT-4 to analyze trends and provide buy/sell signals.

But can AI help traditional markets too? That's Bridgewater's hope, with its new Artificial Investment Associate (IAA) lab planning to analyze the state of financial markets and make predictions for investors to take advantage of.

Previous attempts to do so have yielded somewhat lackluster results — with the Eurekahedge index's AI-driven funds outperforming the broader hedge fund index by around 14 percent over the five years to 2022.

This is mainly required due to the large amount of accurate information related to feeding.

Ralph Kubly, a board member of the Casper Association, believes AI can transform traditional finance — but only if it combines blockchain records with strong standards to ensure the data it feeds into the models is comprehensive and accurate.

For years, he has been advocating that the financial industry adopt the Algorithmic Contract Types Universal Standard, or ACTUS, created after the global financial crisis, caused in part by complex derivatives where no one understood the liabilities or cash flows involved. He believes that standardized data across the chain will be essential to ensure trust and transparency in model results.

“Fundamentally, we believe that without blockchain AI would be pretty lost,” he told the magazine. “Imagine you invest in an AI company, and you're updated every three months about the progress of LLMs, right? If you can't check what you eat in the model, you have no way of knowing if you're making any progress.

It explains blockchain safeguards to prevent companies from compromising their results, and the past shows this […] There is so much money, they lie about what happened.

“AI, without this layer of blockchain verification — what happened, when, where, what was used — I think will be ineffective going forward.

Combining the two will lead to new predictive capabilities, he says.

“Going forward, the hope for AI is that predictive models are becoming more powerful and behavior can be better predicted,” he points to credit scores as an example.

“Used in the right way, AI can lead to very powerful predictive models, which means that people who currently can't get a loan – but they deserve credit – can get a loan. It's something I really like.”

Don't believe the hype: AI's predictive capabilities have so far been shown to be poor at best, and reliable and trustworthy data not recorded on the blockchain can be valuable input for AI analysis.

Also Read:

Real AI use cases in crypto, number 1: The best currency for AI is crypto

Real AI matters in crypto, number 2: AIs can run DAOs.

Real AI matters in crypto, No. 3: Smart contract audits and cyber security

Real AI and crypto use cases, number 4: Fighting AI lies with blockchain

Subscribe

A very engaging read in Blockchain. It is given once a week.

Andrew Fenton

Based in Melbourne, Andrew Fenton is a journalist and editor covering cryptocurrency and blockchain. He has worked as a film journalist for News Corp Australia, SA Wind and national entertainment writer for Melbourne Weekly.

Follow the author @andrewfenton