22 Apr 2023
Will the merger of AI, blockchains and crypto become the next big thing?
With the spotlight on AI applications leading to speculation about the potential of a merger between AI, blockchains, and cryptocurrencies, what are the potential synergies and possibilities?
The convergence of Artificial Intelligence (AI) and its various use cases has been a popular topic in recent months, particularly with the launch of ChatGPT – an AI assistant that can answer questions, write content and provide information on any topic through conversation.
Of course, we have to explore if there are potential synergies between AI, crypto, and blockchain, which we’ll do in this article.
AI is slowly but surely moving towards mass adoption.
While still early, the amount of private investments in AI globally amounted to $93.5B U.S. in 2021, double the amount from 2020.
Then there’s the consumer-facing AI, which has spread like wildfire lately.
This includes generative AI algorithms like MidJourney, Dall-E, Stable Diffusion, AI writers like Jasper and Rytr – and more recently, ChatGPT, which was the first to combine language understanding and pre-trained transformers with GPT-3 and make it available to the public.
ChatGPT has amassed more than 1M users in a remarkable five days, setting a new record for the fastest company to achieve this milestone.
Traditional companies are also hopping on board, with Notion introducing an AI assistant, Google using AI to interpret doctors' handwriting, and Microsoft integrating GPT-3 into their search functions.
All of this shows that the power of AI and machine learning is being harnessed more than ever before. As a result, we believe AI is here to stay and will be an integral and accessible part of our lives in the future, with much more potential for more use cases.
Naturally, the spotlight on AI applications has led to speculation about the potential merger of AI, blockchains, and cryptocurrencies. Could these technologies be combined? What opportunities would that create? What are the leading projects in this area?
In the following few sections, we will explore these questions and more, providing an in-depth look at the possibilities and synergies of combining AI, blockchains, and cryptocurrencies.
Before we get into the potential of the technologies combined, let’s first go over some basic terminologies to understand how AI works.
Artificial Intelligence is a combination of machines that can perform tasks with human-like intelligence —. One of them is a machine playing chess, a self-driving car that can detect its surroundings and navigate without human input.
Next, Machine Learning (ML) is a subset of AI, and it involves giving a computer program access to data and letting it learn from that data. The goal is to enable the program to make decisions or predictions independently without being explicitly programmed.
For example, you can give a machine learning algorithm a bunch of pictures of dogs and cats, and it will learn to recognize the differences between the two types of animals.
By combining AI and ML, you essentially can make a powerful autonomous machine, where the machine learning algorithm might be used to analyze data and identify patterns, while an AI system might be used to make decisions based on that analysis.
For instance, ML algorithms can be used to monitor our sleep schedules, determine what qualifies as ‘sleep’, and track how much time we spend on work.
Such algorithms can then be used by an AI model to create a personalized schedule for each individual that is uniquely tailored to them.
Moreover, this schedule can be adapted and updated as needed, creating an efficient and effective way to keep our lives in check. AI and ML models can thus be incredibly helpful in our day-to-day lives, ensuring that we stay organized and productive.
While these sound exciting, we see that there are some major pitfalls that consumer AIs will run into – particularly, that most AIs, as of now, are black boxes of a combination of machine learning algorithms, which we cannot verify.
Without a trustless way of being able to verify the algorithms, there will be instances in the future where the algorithm will be biased – such as with our example above, in which if a malicious employer uses an AI-powered scheduling system, they can alter the decision-making process to prioritize more work and less personal time.
Of course, this minor example is just the tip of the iceberg; now, imagine if we used black box AI models in more financially oriented decisions.
The implications could be incredibly detrimental to consumers, as AI algorithms could be used to manipulate financial decisions, potentially leading to huge losses in investments.
Malicious actors may even use the AI algorithms to gain access to confidential information or other sensitive data, leading to potential privacy violations. As such, it is crucial that we create a framework for trustless verification of AI algorithms in order to ensure that the decisions made remain fair and unbiased.
Fortunately, this is where the blockchain technology steps in to provide a solution to the security and trustworthiness issues related to AI models. By allowing anyone to verify the data and algorithms used in AI models, using the blockchain can ensure a secure and trustless environment for AI developments. This can then be further refined with the use of cryptocurrencies for protecting against malicious actors, while simultaneously incentivizing good actors that build robust AI models with honest ML algorithms to enable use cases such as data accessibility, democratic algorithms, on-chain AI trading, AI-based information curation and adding of intelligence to NFTs.
Now that we understand why we think that the blockchain and cryptocurrencies will be an essential part of the development of AI, let’s look at the state of the landscape of AI in crypto, and the top projects in this sector.
In this section, we review the state of the current landscape of the AI sector in crypto, followed by looking at the top 5 projects, some complimentary projects to AI, and last but not least, some alpha in the AI space (i.e. Projects that are working on AI but not launched yet) — because after all, the last reason is why you started reading this article, isn’t it?
As of January 1st, 2023, the Artificial Intelligence (AI) sector within the crypto-market comprises only 0.04% of the total crypto industry valuation - a mere $300 million compared to the entire market size of $8 trillion, and with the ever-growing applications of AI, this sector is still relatively nascent.
It's clear to see that the sector has plenty of potential for growth, and naturally, one may ask, "What are the current projects that are working on combining the sheer power of AI with the trustworthiness of blockchains?"
We will discuss this in the following section.
First, let us inspect the metrics that encompass the AI sector in order to properly assess the landscape.
In the chart below, we look at the valuation of the various projects that make up the AI sector.
Here, we see that 5 projects take up 84.7% of the valuation of the sector: Numeraire ($NMR), Fetch.ai ($FET), SingularityNET ($AGIX), Cortex ($CTXC) and PlatON Network ($LAT).
The fact that a small subsector of projects with limited options in the sector is dominating the valuation is not necessarily a bad thing.
This could be indicative of a sector that is still in its infancy, similar to what Uniswap experienced at the beginning of the DeFi revolution. Now, Uniswap has since been joined by a horde of other DEXes, all of which span different chains.
We believe that the AI space is following a similar trajectory, with more and more options becoming available to investors. The sector is growing and evolving, and we anticipate that it will continue to do so in the future.
But how has the performance of AI projects been in the past year?
When assessing the performance of AI projects, it is important to consider the drawdown of the entire industry as a baseline.
To ensure accuracy, we have limited the scope of our evaluation to AI projects with a valuation of over $5M. Projects below this threshold are generally more prone to volatility and do not yet have product-market fit.
The vast majority of projects have performed similarly to the baseline, with a drawdown of approximately 64%. However, there have been some notable exceptions, such as Vectorspace AI ($VXV) at 93%, PlatON Network ($LAT) at 89%, Fetch.ai ($FET) at 81%, and SingularityNET ($AGIX) at 73%.
On the other hand, the Artificial Liquid Intelligence (ALI) token from Alethea stands out as the token with the least drawdown. However, despite this impressive performance, it is important to note the relatively small size of this project at only 15M of market capitalization, as compared to its peers.
Finally, to analyze the performance of AI projects from a month-over-month (MoM) valuation perspective, we took a closer look at the data.
To ensure the accuracy of our results, we only included projects that had a valuation of more than $15M, so that we can reduce the risk of micro-volatility and have a more reliable comparison against the global month-over-month capitalization (MCAP). By looking at these trends, we can get a better understanding of where the industry is headed.
As this chart illustrates, the trends of AI projects are largely in-line with that of the market, with the exception of Numeraire ($NMR) which saw an increase in July before settling just below its January value.
However, in recent months following the launch of ChatGPT, these projects have been breaking out on their own, suggesting there is strong traction in the sector. This is a promising sign for the future of AI technologies, and we can only anticipate further progress in the coming months.
Now, let’s look at the top 5 projects in terms of valuation, and what they are actually solving or trying to build in terms of AI. Here, we take a look at Numeraire ($NMR), Fetch.ai ($FET), SingularityNET ($AGIX), Cortex ($CTXC) and PlatON Network ($LAT).
For these examples, we won’t be going in depth on the nitty gritty of the individual projects’ tokenomics, but aim to find out the answers to these questions:
- What is the chain trying to solve in the AI space?
- What blockchain is it built on?
- (If any) What interesting projects are built using the chain?
After we find out the answers to these questions, we will get a sense of where the holes are in developments in AI and Blockchain, and from there, we then assess other projects that are complementary, and also still in development.
Numeraire, a hedge-fund like software that was built on the Ethereum blockchain in 2017, enables users to trade stocks on the Ethereum blockchain without the influence of human emotion.
Powered by a "meta-model" artificial intelligence aggregation model, Numeraire leverages the blockchain technology to enable users to stake cryptocurrency on predictions made by a global network of anonymous data scientists who are using artificial intelligence. This ensures that strategy secrets from other hedge funds and individual traders are not compromised.
In addition, Numeraire has two applications that serve different types of traders:
- Signals, which is a channel for submitting stock market techniques that promote a specific stock or trading style
- Tournament, which is a weekly competition in which participants submit trading algorithms based on stock market data and make predictions about a dataset. At the end of the week, participants who make correct predictions are rewarded with NMR tokens, while those who make incorrect predictions have their NMR tokens burnt.
Based on the Cosmos-SDK and CosmWasm, Fetch.ai is a platform that uses AI and ML to automate marketplaces, allowing for near-frictionless economic transactions that happen at a high rate, and they aim to achieve this by using four distinct elements:
- Agent Framework, which supplies modular and reusable components for developing multi-agent systems. Agents act on behalf of people, devices, and organizations by collecting data and making predictions. For example, agents can forecast how much fuel to use depending on traffic and weather. They use algorithms to find the best solution and can also learn from past mistakes.
- Open Economic Framework, which offers search and discovery tools for agents to find each other and send messages.
- Agent Metropolis, which is a collection of smart contracts that uses WebAssembly to keep track of agreements and provide services for agents.
- Fetch.AI Blockchain, which uses cryptography and game theory to enable secure, censorship-resistant consensus and fast chain-syncing.
Here, AI and Machine Learning are used to evaluate trust and market intelligence, and Fetch.AI has improved efficiency by introducing solutions to everyday problems, such as creating AI ‘agents’ to counter DeFi impermanent losses. These agents can help liquidity providers on platforms like Uniswap and PancakeSwap by setting stop-loss triggers and providing a dashboard to monitor and update the agents.
Projects that are powered by Fetch includes, but not limited to:
- Axim, which is an innovative platform that enables businesses to seamlessly connect their data silos, enabling them to gain valuable insights to help optimize their business functions, whilst safeguarding the security and privacy of their data. Axim's advanced machine learning capabilities further allow businesses to gain a deeper understanding of their data, uncover hidden patterns and correlations, and use the collected data to make better decisions for their business.
- CoLearn, which is an innovative blockchain-mediated collective learning system that facilitates collaboration between multiple stakeholders to construct a shared machine learning model. It eliminates the need for data to be contained in siloes, thus allowing machine learning initiatives to be undertaken without requiring trust between participants.
Built on the Ethereum blockchain, SingularityNET is trying to ease the curve in accessing the using AI technologies, as it can be expensive and complex. They are doing this by creating a decentralized marketplace that allows developers to easily publish and monetize their AI algorithms with a public API and decentralized marketplace.
AI agents run nodes on the platform and can autonomously interact with other agents to exchange data and supply results. Agents are rated on a scale from 0 to 5 based on their performance, consumer ratings, financial transaction trails, and escrow patterns. Any agent with a rating of 2 or higher can take part in on-chain governance.
The platform also tracks interactions to become more efficient, with the ultimate goal of creating a reflection of human neural networks, with the components that allow this to function being:
- Registry, which is a smart contract that forces AI developers to explain their services, so participants can check if they are compliant.
- Multi-party escrow, which is a smart contract that governs transactions in a marketplace using atomic channels for fast settlements.
- Daemon, which is a proxy that checks if escrow conditions are met and validates requests off-chain.
Also, SingularityNET has enabled some groundbreaking AI projects that, such as:
- SingularityDAO, which is revolutionizing the concept of crypto asset management by allowing anyone to take part with a multitude of various tools, such as the DynaSets which are risk-managed AI portfolios built with data science and trustless smart contracts.
- Awakening Health, the creators of Grace, an AI robot specifically designed for elder care. Grace has been deployed in a Montreal nursing home in 2022 to provide seniors with companionship and support.
These are just two of the many projects that have been made possible with the help of SingularityNET, proving the platform’s immense potential for the future.
Built on the Ethereum blockchain, Cortex aims to provide users with access to the latest machine-learning models through the use of smart contracts, by creating a marketplace on the Cortex blockchain.
Cortex also wants to create a machine-learning platform where users can post tasks and submit AI decentralized applications, or also known as AI dApps by allowing developers to add artificial intelligence (AI) to smart contracts, through the use of:
- Smart AI contract, that lets users write machine-learning programs on top of the blockchain.
- A virtual machine (CVM), that is compatible with EVM, where researchers can upload data models, with training, to the storage layer, which can then later be inferred by developers via modules.
Cortex also features a reward system that encourages people to work together on models, where results are agreed upon by consensus, and models can be trained off-chain.
PlatON is building a decentralized AI network and global brain to make AI available to anyone. They are aiming to achieve this through their three-tier network model, which includes:
- A consensus network that enables smart contracts with extra privacy-preserving protocols such as homomorphic encryption, zero-knowledge proof, and TEE that is overlaid to protect data and computations on the chain.
- Privacy-preserving computation network, where data and algorithms are kept locally and computations are done securely and a decentralized market for data, algorithms, and computing power.
- (Conceptual Stage) Decentralized AI collaboration network, where they are allowing AI agents from projects such as SingularityNET, Botchain to collaborate, and enable autonomous AI organizations.
If the AI industry were to increase in value, it wouldn’t just solely benefit AI. For example, because AI requires massive amounts of data, we would expect data projects to increase in value as well.
In this section, we look at projects related to AI and other early projects that are building in the AI space.
First, we look at some of the complementary projects to the AI sector, as the AI sector uses blockchain for trustless data sources, like Oracles do for DeFi.
Built on Ethereum, Ocean Protocol aims to unlock the potential of data for use in Web 3.0, by connecting data providers with entities who are willing to buy or access the data.
Data providers are paid in OCEAN tokens for allowing access to their datasets. The data itself is stored off-chain and the access rights to it are represented on-chain using OCEAN tokens. The two main players on the platform are the sellers (providers) and buyers (consumers).
Providers can be any individual or business who possess a dataset that they want to monetize access to. Consumers are typically businesses of all sizes who want to access datasets for a variety of purposes.
The platform is being used for various applications, such as in the automotive and healthcare industries, and Ocean Protocol has an active community, aided by the community funding for software development, incentives, and grants to encourage users to supply data to the network.
Next, let’s take a look at some of the AI projects that are early in the development stage and have not gone mainnet yet, or are not listed on token aggregators such as Coingecko or CMC.
Building on Polkadot, Bittensor is an open-source protocol which aims to create a decentralized AI network that is more efficient, transparent, and cost-effective than current options by powering a decentralized, blockchain-based machine learning network and allowing various machine learning models to collaborate.
The Bittensor Network uses a carefully designed consensus mechanism that is subject to change as the network scales, with the main goal to deter malicious activity, incentivize honest assessments, and promote the interaction of high-value peers in the network.
Servers receive two separate scores that determine their earnings, and Validators are subject to bonding and score supply conditions.
Because this project is relatively new, there are currently no external projects building using BitTensor that are finalized, as of yet.
Now that we’ve seen what some of the use-cases (the rest are more or less the same) of applications of AI in blockchains, let’s figure out where we think this sector is headed towards.
For a fact, we know the AI space is definitely not a sector to shy away from, as the technology is getting so advanced that we may indeed enter an era of superhistory, and as the amount of data gets much more advanced, it is definitely wise for AI projects to look upon the blockchain for a more trustless way of deployment and extraction of datasets.
In fact, there has been much more money thrown into AI products and use-cases — which begs the question: what other use-cases are there for AI and blockchains, and what are the limitations of the blockchain as of current that is not allowing for these use-cases to appear?
Vast amount of datasets: One of the biggest reasons there aren’t much AI applications on the blockchain is because AI requires vast amounts of data and runs with algorithms. While having a blockchain or ledger that is filled with strings not requiring computations is easy, and also not very heavy in terms of computations, this is not the case for AI applications.
With heavy computations, this will introduce higher fees for computing — and if we were to be in the bull market where trading NFTs on Ethereum costing us rent money, imagine running multiple complex ML algorithms just so that we can do one thing with an AI.
This brings us to the next point.
Lack of modular blockchains: So at this point, you’re like, how do we fix it? Well the answer is not as simple as it seems. I’d suggest that one of the ways this can be fixed is via having modular blockchains that are retrofitted for the purpose it serves, and not by making AI applications on top of monolithic blockchains.
In our example where we consider AIs, ML algorithms are not one and the same, we should also be able to have in a similar fashion, blockchains that are suited only for creating and processing AI models, ML algorithms, and much more — and to have all of them interoperable.
One other last thing that we want to point out: We’ve also noticed that most projects could not figure out how to best reward users to use AI with tokenomics, which we feel is also one of the limiting factors in getting mass adoption for projects.
Smart Routing: If and only if the problems of throughput and storage are fixed, then can we look at other interesting use-cases for AI, such as for example, requesting an AI “chatbot” to help you trade X to Y, with your input being conversational. This is how I’d imagine this to be like:
- You: Can you trade for me 10 X to 10 Y, and make sure that I get a minimum of $100 Y
- Chatbot: Sure, according to my predictions, I will set the price to sell X at $xxx, and it is expected to be sold on DDMM. Do you want to continue?
Create automated smart contracts: Or, you could also have a conversational way of creating smart contracts, such as just stating that you want an Escrow between X and Y with your parameters, and the AI bot will create these contracts for you with the updated best practices that is aggregated from multiple blockchains and in any language you want.
Thus, in our opinion, we think that the AI sector is one sector that is nascent.
We think that while current blockchain implementations of AI are interesting, there is definitely much more potential that has yet to be unlocked within this intersection — and even use cases we might not imagine would be possible until there are more advancements in the blockchain space.
For this reason, we will update the article every few months with the latest news about this sector in this intersection.
The Flagship AI Vault
Invest in the entire Crypto AI sector in one transaction!Learn more!
Disclaimer: Nothing on this site should be construed as a financial investment recommendation. It’s important to understand that investing is a high-risk activity. Investments expose money to potential loss.