AI and Blockchain Disruption: Unveiling Perfect Synergy Use Cases

Where does blockchain meet AI?

Can blockchain be perceived as disruptive as artificial intelligence? Will rapid AI developments cause VCs to forget entirely about Web3? And does the blockchain world need AI only because of the energy created through the huge hype?

Despite the uneven attention the two technologies have received over the past few months, their disruptive power is quite comparable. However, there is no competition between them. Quite the opposite is true. We believe AI and blockchain can and should coexist and cooperate.

This research report reveals why and explores how.

Source: https://medium.com/@scalingx/exploring-the-horizon-the-convergence-of-ai-crypto-38086af96987

Our initial research for this report led us in a distinct direction. We quickly realized that it’s not about assessing whether or not combining AI and blockchain makes sense. The powerful synergy between the two became obvious from the very first industry report we analyzed.

The data collected from the survey respondents (including people familiar with AI, blockchain, or both) further confirmed this assessment. This convergence’s transformative potential became apparent in real-world challenges and its ability to improve businesses.

Hence, this paper focuses on when and where the combination of AI and blockchain can bring the most value. 

Is it particularly useful for a specific industry? Can it enhance or create a unique business model? Or are there internal company processes that are particularly “revolutionizable” by the convergence?

Stay with us as we dig deep into these critical entrepreneurial considerations.

Everything you’ve wanted to know about AI (but were afraid to ask?)

Before we embark on a journey to explore the most useful, promising, and popular synergies between AI and blockchain, let’s clarify the term that you are probably less familiar with: AI. We believe you already know a thing or two about blockchain. 

What is artificial intelligence, anyway?

Artificial intelligence is a field of computer science that started in the late 1950s. It is dedicated to creating intelligent machines that can think, learn, and act autonomously – similar to humans. The idea is to make computers smart enough to carry out simple tasks that usually require human intelligence, like understanding language, recognizing patterns, or making decisions.

This is how AI operates in a nutshell: AI gathers data from various sources and preprocesses it for analysis and selection of appropriate algorithms for the task at hand. It trains machine learning models if needed, makes predictions or decisions based on the analyzed data, and incorporates feedback to improve performance over time. This enables AI systems to emulate human-like intelligence and autonomously perform tasks ranging from image recognition to recommendation systems and self-driving cars.

ChatGPT isn’t all that artificial intelligence has to offer

ChatGPT is a specific type of AI, like a powerful tool in a toolbox. It’s a large language model (LLM) that excels at generating human-quality text. But artificial intelligence is much broader. It encompasses everything from self-driving cars that learn to navigate to medical diagnosis tools that identify patterns in patient data. 

What’s the difference between artificial intelligence and machine learning? 

This is a common question, so how can we stop confusing the two? AI is responsible for achieving the overarching goal of creating intelligent machines, and machine learning is a technique used to fulfill it. Think of it this way: AI is the destination, and machine learning is one of the vehicles that can take us there. 

Source: https://www.mlguru.ai/Learn/concepts-deep-learning

Where does blockchain technology fit into the picture? 

Blockchain technology serves as the underlying technology for the infrastructure required for many AI tools. At the same time, blockchain adds value to AI projects and AI-driven tools by providing an additional security layer, transparency, decentralization, automated mining processes, data analysis, secure storage, and data management. 

In simpler terms, blockchain creates a secure, transparent, and decentralized infrastructure needed for AI tools to function effectively in the world of open-source projects. And we’re happy to elaborate more on this in the following sections. 

If you’d like to make yourself familiar with some of the critical terms surrounding blockchain x AI intersections before we dive in, please check out our short glossary.

Web3 AI x blockchain market

Let’s start by reviewing the current state of the AI x blockchain market, giving you an overview of the projects we’ll discuss here.

Our first research step was an initial screening of the AI part of the Web3 market. We analyzed 107 companies that fit this description. We then categorized them into the following 13 sectors (this is explained in more detail in the “AI x blockchain – synergies overview” section):

We broke the categories down further based on the value projects bring to the market. We identified the following three main types:

• Suppliers – Web3 projects that provide other companies with resources necessary for creating or maintaining AI tools. Good examples are Akash and OpSec, which use DePIN to deliver computational power, which is essential for AI algorithms.
• Providers – Web3 projects that provide other companies with AI-based tools. For example, Fetch.ai and Autonolas facilitate decentralized crypto-AI agents. 
• Users – Web3 projects that implement AI solutions to improve already existing blockchain-based initiatives. For instance, Web3 games like Sleepless.ai and Sidus use artificial intelligence to optimize in-game experiences.

The obvious question now is: How are the Web3 projects distributed among the three types?  

It’s safe to say that it’s relatively balanced. The most universal group are the providers, and they’re also the largest, with the abject sides of the spectrum almost identical in size. Note that some companies supply AI tools with data, computational power, etc., and provide solutions on their own. That’s why the sum of the projects in the below graph exceeds 107.

The last part of this initial analysis focused on the services provided by such companies. We investigated the specific utilities Web3 AI projects offer for both users and businesses, starting with the aforementioned decentralized computing, moving through decentralized generative AI models, and ending with smart contract auditing and generation.

The following graph shows the number of projects in each service category.

And if you’d like to see more concrete examples of projects operating in specific niches, here is a blockchain x AI sector map for you:

You must be keen to grasp what these niches, utilities, and numbers mean for you as an entrepreneur. Well, now that we are on the same page regarding the projects that capitalize on or provide the most value from AI x blockchain convergence, we can dive into specific use cases of various synergies.

AI x blockchain – synergies overview

To pinpoint the potential impact, it’s best to look at this union from two different angles:

1. How does blockchain already help AI? Use cases related to AI problems that blockchain technology addresses, e.g., the lack of computational power to run AI algorithms → utilizing the unused GPU power from personal computers thanks to DePIN technology.
2. How does AI already help blockchain? Use cases related to blockchain problems that artificial intelligence addresses, e.g., vulnerabilities in smart contracts → automated AI-based smart contract auditing.

We also included an analysis of specific business implications (e.g., the changes in revenue streams, cost structure, and accessible markets thanks to implementing AI x blockchain synergies). You’ll find it all in the following three sections of the report. 

How does blockchain already help AI?

Decentralized computing power

AI is currently not in a position to overtake the world. It’s way more focused on ordinary, playful, or routine tasks. Like this one: finding enough computational power to create yet another image based on the “Dwayne The Rock Johnson as a Rock” prompt.

Regardless of the task, AI algorithms always need graphic processing units (GPUs) to perform them. The sudden increase in demand for advanced chips such as Nvidia’s H100 has caused a shortage, driving up the costs of such computational power. Therefore, it may become impossible for individuals or smaller businesses to use AI at full scale, leaving it in the hands of well-funded corporations.

That’s where DePIN, a blockchain-based technology covered in our previous report, comes in, along with projects like Akash, OpSec, and the Golem Network. They serve as a type of Uber or Airbnb for computational power and create a two-sided marketplace. Companies needing GPUs can “rent” them from businesses and individuals who share them in exchange for token-based incentives.

Such an approach significantly decreases the cost of computational power. Look at a simple comparison of Akash pricing with the largest representatives of a non-Web3 computing market:

It’s also important to mention that the DePIN x AI intersection benefits aren’t limited to businesses or people using high-end GPUs such as Nvidia A100s. The recently well-funded GPU network io.net ($30 million in Series A) allows the sharing of the power coming from the Apple’s M1 or M2 chips. 

So, if you’re currently reading this report on your Macbook, you may consider an additional source of income. 

Security risks in AI

As artificial intelligence systems become increasingly sophisticated and integrated into our lives, they also introduce potential security risks that need careful consideration. 

• One major concern is the trust problem surrounding AI data. How can we ensure the data used to train these systems is accurate, unbiased, and ethically sourced?
• Additionally, the centralized nature of many AI platforms raises a red flag about the potential misuse of data and lack of control for individuals. 

Therefore, safeguarding privacy and establishing security measures are crucial to maintaining trust and ensuring the ethical and responsible use of AI technology.

The fundamental trust problem for AI data

AI adoption expands at an extraordinary pace – especially when fueled by Big Data, the resource that enables AI to learn, become smarter, and unlock insights from an ocean of information.

However, as AI advances, the trust problem grows proportionately. Discerning between genuine and AI-generated content becomes increasingly difficult. Generative AI, while capable of stunning creativity, lowers the barrier for misinformation and deepfakes, demanding a renewed focus on content authentication. For example, an AI system trained primarily on images of white men performs poorly when identifying people of color or women, leading to biases and discrimination.

This erosion of trust fuels the need for solutions, and blockchain can offer some. Take these four as examples:

1. For those who are already tired of fake news and misleading AI-generated media (all of us?), Fox Corporation has taken a bold step. The news company launched Verify, a tool built on Polygon where publishers establish undeniable proof of content ownership. Each piece is secured on the blockchain, empowering consumers to easily identify trustworthy sources using the Verify Tool.

2. Teams from the Arweave blockchain and the provenance layer, Irys (formerly Bundlr), have followed suit and developed the Digital Content Provenance Record (DCPR) standard. This innovative solution utilizes the Arweave blockchain to timestamp and verify digital content. By providing reliable metadata, DCPR empowers users to assess the credibility of online information with greater confidence.

3. Similarly, skincare firm Clarins combats counterfeit ingredients with cutting-edge blockchain verification. Partnering with NeuroChain, an ML-based consensus blockchain and Community, they built the Clarins T.R.U.S.T. solution. Suppliers log ingredients into the system, while the blockchain provides proof of authenticity, ensuring only genuine materials enter their supply chain.

A scalability test carried out by Clarins produced the following results: They were able to open 160,000 pages of their website/software interface within 1 minute without any degradation and without any negative effects on response times or availability.

4. The last example comes from the world of global enterprises. The Geoscience Team of the Abu Dhabi National Oil Company (ADNOC) used AI to augment and accelerate the visual photography process. ADNOC has partnered with software giant IBM to pilot blockchain-powered transaction management for its oil and gas operations. This combined solution helps the organization automatically track, validate, and execute transactions throughout the supply chain.

Centralization risk of AI
While open-source AI promises democratization, a closer look reveals a troubling reality: the control seems to remain in the hands of a few tech leaders. Google, Amazon, and Microsoft appear to dominate AI development by concentrating the necessary data.

Centralized data storage raises the following concerns:

• A single point of vulnerability: breaches at large data holders expose vast amounts of information.
• Lack of transparency: users often have little control or visibility over how their data is used.

To combat these challenges, companies are leveraging blockchain and AI to revolutionize how data is stored, managed, and used, with profound implications for healthcare and beyond. One such example is the partnership between Akash Network and Solve.Care, which is using these technologies to change how patient data is handled.

Due to current industry regulations, patient data must be stored in environments such as Amazon Web Services (AWS) and Google Cloud Platform (GCP). Under this system, patients lack sovereignty over their own data. Additionally, there is no guarantee of digital permanence if data is lost.

Akash Network and Solve.Care’s partnership allows patients to own and control their health information securely through blockchain technology. Akash provides the computing power necessary to deploy Solve.Care’s proprietary node infrastructure (Care.Nodes) to patients globally. As a distributed network with compute providers worldwide, Akash allows Solve.Care to access computing resources without permission in the regions where their enterprise clients and patients reside and receive care.

Another inspiring example of this approach is Thermaiscan, which explores blockchain to improve healthcare data security:

However, such a transition doesn’t happen without challenges. Especially in an “analog” industry like healthcare. As stated by Arby, “Many countries and healthcare systems still use paper and pens. Moving into digital systems powered by blockchain will require them to make a “jump” compared to some developed western countries that use standard databases provided and hosted by big cloud server companies.

But according to Thermaiscan, it’s still achievable. Especially in emerging markets: “If we take Swedish healthcare providers, they still use hard disks to store data – since they don’t want to contract international cloud providers. However, low and middle-income countries where Thermaiscan has been engaged have a higher adoption and acceptance rate of new technologies. There, it’s way easier to find right-minded, healthcare-oriented people and convince them to at least start testing such solutions.

Healthcare is by far not the only industry experimenting with this synergy of technologies. In a joint effort, Fetch.ai (partnering with peaq), Deutsche Telekom, and Bosch use the AI x blockchain convergence to transform IoT devices.

At the heart of their collaboration is the Bosch XDK110 sensor kit, which serves as a self-contained data acquisition solution and is capable of measuring environmental factors (such as seismic activity, humidity, pressure, etc.) for various applications. Each sensor has a secure, blockchain-powered identity that enables the identification of a machine by giving it a unique identifier generated by the blockchain protocol of peaq ID. Fetch.ai’s AI agent allows the sensor to learn, function independently, and even earn revenue by selling its data. At the same time,  the sensor data can improve traffic management or air quality monitoring. You’ll learn more about AI agents in the “Crypto Infrastructure for Agents” section.

Machine learning privacy

AI systems are hungry for vast datasets, regardless of whether they contain sensitive information or not.

Let’s stick to the example of health data because privacy is particularly critical here. The personal genomics company Nebula Genomics teamed up with Web3 company Oasis Labs to give people more control over their genetic information. They use a new tool called Parcel which is designed to let users decide who can access their data based on a complete history of information usage stored on the blockchain. This way, patients can see who has looked at their genetic information and stop such activities whenever they want. 

In another example, iExec, a marketplace for AI models, leverages blockchain technology and Intel’s SGX hardware to ensure secure sharing and utilization of AI models. When these models are rented on iExec, they remain encrypted until reaching a secure environment provided by Intel SGX. This protects the intellectual property of AI developers and promotes trust within the AI ecosystem, where concerns about model theft and misuse are prevalent.

AI data decentralization

Research on AI is flourishing. Based on the results derived from the EBSCO database, since 2021, over 1,300,000 academic papers have been published on this topic. That’s more than throughout the entire last decade (1,100,000). 

Does the quality meet the quantity, though? Not quite. Research is done repetitively by a small group of authors, and there is a growing number of flashy journals and biased reviews. This suggests that there’s a need for change. Another argument is that “data” remains the most frequently cited challenge in discussions surrounding AI, as our survey indicates (you can see it from the following graphic).

That’s where decentralized AI research fits in. It’s an innovation highly focused on incentivizing data sharing. Projects like Bittensor and Allora use the weighing system to encourage participants to co-create ML models by focusing on the most useful data. This approach is based on analyzing a specific task (e.g., predicting the weather), weighing the importance of specific data when building a model to conduct it, and – most importantly – incentivizing participants through economic rewards to share the most vital data. 

If you read our previous report on DePIN business opportunities, you discovered that decentralized AI research is often combined with DePIN mechanisms. A simple example is Grass, which pays its users to share data scraped from the public web. This is essential for even basic AI tools such as ChatGPT.

The casual gaming company NeuroMesh invests significant resources in building distributed training networks. Its executives believe in the potential that incentivization brings to building AI models.

Similarly, the Chief Scientist of the AI-powered oracle ORA Protocol sees it as one of the most powerful convergences of both technologies. 

It’s also worth mentioning that, in the case of AI research, decentralization goes beyond expanding the number of data sources through incentivization. Cere Network takes it one step further and adds a layer of shared storage. 

A single point of failure is perceived as something blockchain can successfully address; therefore, securing AI data by decentralizing serves as a viable option. It’s another intersection between AI and DePIN – Cere Network compensates its users for providing storage in the same way as Filecoin and Arveawe. 

All these factors contribute to the shared view among people familiar with blockchain, AI, or either of these areas that decentralized AI is the synergy with the highest potential for the future.

Crypto infrastructure for AI agents

A dystopian vision of an army of autonomous AI-powered individuals is already coming to fruition. And blockchain can make it even more frightening – or promising.

A so-called AI agent is a software program that acts on your behalf based on provided data. The data can come from a variety of sources, including manual input from humans, sensors, or cameras. An AI agent is designed to make an informed decision and act according to specifically defined requirements.

It may sound a bit vague and complicated, but examples of AI agents are already surrounding us. Innovations such as autonomous vehicles, virtual assistants, or even good old chatbots in the form of ChatGPT all rely on AI agent principles. 

However, despite a vast set of use cases already present, such solutions lack one important aspect to become truly powerful: built-in economics. AI agents are unable to open or use a bank account and spend money on your behalf. Let’s be honest; they’d immediately fail the KYC process. And even if we provided them with the necessary permissions (you can imagine the endless back-and-forth with the bank or its chatbot AI agent), such a tandem would likely become cumbersome, inefficient, and dangerous for your personal finances.

Providing crypto-AI agents with a blockchain infrastructure adds an economic layer to their work while also addressing security concerns. 

One of the most promising examples of this intersection is Fetch.ai. On the one hand, it enables developers to build secure and token-governed autonomous agents with the Fetch.ai software. On the other hand, it provides a marketplace called agentverse, where AI deputies can transact, negotiate, and chain with each other, forming “working groups.” 

As a result, AI agents become more autonomous and accomplish tasks in a collective manner. When an agent requires data or a specific activity, it can purchase it from another participant in the agentverse. At the same time, the entire network is still governed by humans and Fetch.ai token holders. They need to ensure their agents focus on assigned duties, such as building real-time EV charging station maps (an actual use case), instead of taking over the entire world.

Ocean Protocol also explores this particular use case. Ocean sees blockchain as a decentralized substrate to support sovereign AI agents, i.e., “AI DAOs”:

Another project that utilizes the AI DAO concept is Autonolas. Here, a group of OLAS token holders facilitates governance over the AI agents, enabling more complex cooperation methods between contracts. By doing so, the project helps businesses and developers build an army of autonomous agents that work collectively towards an outlined goal. Companies, teams, people… who needs them anyway? Scrap that; we never said that!

Addressing the AI carbon footprint

It’s estimated that data center energy consumption in Europe will grow by 28% by 2030. Needless to say, AI is a major contributor to this massive increase. To put things in perspective, training ChatGPT-3 generated 552 metric tonnes of carbon, which is equivalent to driving 112 petrol-powered cars for a year.