AI and blockchain technologies: a powerful synergy!
Two huge and hot techs are literally fusing together into new, sometimes brain wrecking, applications.
I am starting an Omdena Challenge in a couple of days (July 10) to explore the deep fusion (‘Fusion) of AI and blockchain technologies (‘Blockchain’) with participants from around the world.
I also invite you to my Fusion group on LinkedIn ‘Blockchain 4.0 — the Fusion of Web3, AI, IoT & Metaverse’.
Here, I’ll give a brief overview of what the Fusion is. You are invited to participate if you feel like learning more about it. I recommend reading section ‘Combining Blockchain and AI‘ in the excellent paper ‘Blockchain: The Next Breakthrough in the Rapid Progress of AI’.
To keep you motivated and amused, and in perfect alignment with this article, I’ve created a few funny or utopian images of the Fusion using Langchain and Stable Diffusion from Huggingface. In fact, I asked ChatGPT to generate the prompt for the diffusion model.
The Fusion is not just using AI to help the blockchain world by, say, detecting frauds or pump and dump schemes, or using Large Language Models here and there. First of all, it’s not about just cryptocurrencies at all. In fact, even if all cryptos were made illegal by all governments in the world, Blockchain would stay. It is a mistake to assimilate Blockchain with cryptos.
No, it goes much deeper than that as you’re going to see, and it goes both ways. I mean, AI needs Blockchain (“What? Sacrilege!”) as I explain below. And of course, everything needs AI now.
In fact, it’s a triple fusion. Let’s just say a few words about it.
The triple fusion between AI, Blockchain and IoT
These technologies are accelerating as even a blind man can see, and they need each other more and more because they fix/address each other’s issues and needs. I won’t talk much about IoT here, but it’s going to invade our lives with sensors. And even IoT needs Blockchain! Surprise!
The IEEE has chosen a blockchain to store IoT data (IOTA: feeless and free). In Blockchain Technology: A Smart and Efficient Way for Securing IoT Communication (2021), we can read “IoT appliances and devices are the boom in the technology but the biggest concern in IoT devices is its security. A lot of attacks are discussed on IoT. Existing technologies are not enough to secure the IoT devices completely. Blockchain IoT is the mechanism of using security feature of Blockchain in IoT to make it more secure and vulnerable to communication attacks.“
Keep the bolded comment in mind. You’ll understand it later. For now, try to imagine the consequences of that statement already when IoT gets pushed into everything.
Why does AI need blockchain?
In this article from IBM, Dealing with the AI data dilemma, we learn that AI suffers from several fundamental issues:
AI models’ outcome can be altered in various ways by simply manipulating the data, maliciously or unintentionally, resulting in a general lack of trust in anything AI. It is actually quite easy and amusing to add a few pixels of noise to an image of a bale of hay, and have a model say it’s a cat.
Data quality: AI is highly sensitive to data, making it critical that it corresponds to the underlying reality.
Data accuracy: outdated or delayed data can render AI predictions inaccurate.
Data scarcity: no one but Big Tech has huge datasets on people, and even they don’t have access to everything.
Data privacy: some data cannot be released online (financial transactions, medical records).
Big Tech, with their huge data processing centres and limitless resources, will dominate the future, and be able to manipulate opinions, elections, trends, prices etc.
Current systems are centralized (at the control level) and therefore have central points of failure, and can be hacked or abused.
Blockchain solves or improves these issues thanks to:
Immutable, censorship resistant, ledgers, accessible by everyone, in real-time.
Consensus mechanisms preventing fraudsters: trust in the data will increase, and therefore in AI services previously mentioned. This will allow robust, permission-based governance (ie algorithmic decision making).
Blockchain protocols such as Ocean allow users to not only share their (encrypted) data, but also receive compensation when AI algorithms use it. This can greatly increase the amount of reliable data available, and decrease Big Tech quasi-monopoly on data.
Decentralization removes central points of failure
Smart contracts, a form of decentralization automation, i.e. immutable code running on the blockchain infrastructures, increase trust by acting as an incorruptible (hm, if properly coded) transaction layer for AI services.
The capability to allow even untrusted parties to communicate on a P2P network and collaborate on a common data structure will increase the integrity and reliability of the data, and therefore the applications themselves.
Decentralized applications, and decentralized (autonomous) organizations (DAO, also called ‘algorithmic governance’) with automated decision taking (Blockchain 4.0).
Why does Blockchain need AI?
“ The problem is that we are generating far more data than we can digest [without AI], much less used for decision-making” (The Intersection Between AI And Blockchain Technology – Industries Of Tomorrow (Forbes))
The image below shows how deeply AI will impact every corner of our lives, every sector of the industry. The biggest tech companies have invested enormous capital and manpower to develop efficient algorithms, and if recent history is an indication, it will only accelerate.
Blockchain reaches and improves many sectors because of its innate ability to provide secure, immutable and decentralized solutions. However, Blockchain is not designed to process huge amounts of data, but AI algorithms can. In fact, “global data volume was estimated to exceed 40 exabytes in 2020, and even pre-deep learning algorithms cannot cope with the growth of data volume” (NextGeneration Enterprise Artificial Intelligence Systems).
AI perfectly addresses these needs. It will foster Blockchain expansion by providing the following services:
Extraction of patterns from huge databases (data, text and images): behavior, trends, personal preferences of specific groups, (ir)relevance of information, identification of risks, threats, scams, opportunities, disinformation and hatred…
Large Language Models in particular will bring tremendous advantages by being able to interact directly with humans and extract information from billions of pages of regulations, scientific studies etc.
Robust and automatic code generation, debug and testing (an issue with smart contracts at the moment)
Capacity to predict prices, cycles, trends, users needs, natural disasters
Objectivity, elimination of bias
Lives improvement: people will adapt and find new ways to work more efficiently
Creativity: thanks to generative AI, algorithms are already able to create art, music, solve difficult problems (protein folding), or maths problems. For instance, computational creativity can produce complex designs for the aeronautical industry.
Security: anticipating / preventing cyberattacks, detecting and correcting bugs and malicious patterns in smart contracts (hacking is currently a major source of loss of funds). AI can analyze and learn from previous hacking patterns but also imagine millions of new ones to robustify smart contracts like no human team can.
Autonomous decision making: Blockchain 4.0, automated decentralized governance (DAO) cannot happen without AI
Generating metrics: to help optimize the networks for instance
Enabling adaptive consensus mechanisms.
Some researchers propose to replace computation-heavy consensus algorithms that require trillions of hashes, with useful computation by designing the block verification and consensus mechanism as a deep reinforcement learning iteration process, to design the next generation of public blockchain networks.
Potential horizontal target markets
Blockchain and AI will impact many horizontal markets, if not all. There are obvious direct benefits from reliable code generation, debugging and testing, data sharing, decentralized computing and applications thanks to smart contracts, with increased security and reliability and trust.
One would not think about cybersecurity, but “threats are increasing at an alarming rate, and the traditional methods of securing systems are quickly becoming irrelevant to new attack vectors”. AI, detecting and responding to threats, with Blockchain, ensuring the security and integrity of data, can create more secure and efficient cybersecurity systems.
Another sector that is not well publicized is the new databases that enhance existing capabilities. For instance, BigchainDB “is software that has database properties (e.g. high transaction rate, low latency, indexing & querying of structured data), enhanced by blockchain properties (e.g. decentralization, immutability, owner-controlled assets)“, thereby improving the reliability of AI algorithms.
With Space and Time the future of databases and analytics is already here. It is an AI-powered decentralized data warehouse which translates any “question about data from any source database (including from blockchains) or data warehouse with enterprise-grade security, privacy, and access controls. Generate queries, pipelines, and dashboards automatically with natural language prompts enabled by OpenAI”.
Their ‘proof of SQL’ allows trustless smart contract to execute tamperproof SQL, by connecting data and analytics directly to smart contracts with cryptographic guarantees.
Tokenization of assets, i.e. representing real assets by a digital form, a token, is a huge market since everything is ‘going digital’ now. Smart contracts will regulate and protect their use. The legal grounds are still shaky, but, once established, it will greatly improve the tracing of all assets, and therefore protect them against frauds and theft thanks to unbreakable immutable proofs of ownership. AI will contribute to the security thanks to Computer Vision algorithms able to analyze biometrics, recognize a face, much more reliably than, say, current visual inspections of a signature.
A particular form of tokens, NFTs (Non Fungible Tokens) are digital creations that cannot be split into parts or duplicated. For that reason, they are very popular in the digital art industry because they can protect artists’ digital creations, paintings, music. AI can help predicting their prices, like for stocks or cryptos’ Automated Market Maker algorithms, but it can also create NFTs thanks to the latest algorithms based on Stable Diffusion (DALL-E from OpenAI, or Binance’s “Bicasso”). NFT may seem futile at the moment, but the technologies involved are extremely powerful. In my view, it’s a rehearsal of things to come (a bit like bitcoin was 15 years ago). NFTs foster creativity and allow anyone to make and sell their own creations.
I briefly mentioned decentralized autonomous organisations (DAOs). AI will drive Blockchain 4.0 into many sectors by allowing automation and optimization at every level, ie. potentially creating an optimal digital society without human intervention. Based on specified criteria and analysis of historical and real-time data, AI can optimize the design of the smart contracts used by a DAO, improve in real-time the governance using voting patterns or members activity, filter such members according to desired criteria (skills, reputation etc), manage its crypto assets (portfolio optimization, risk management, trading etc), and, as always, prevent fraudulent behaviors. This may seem far-fetched or utopian, but the potential is there now.
I could go into the details for many more horizontal markets, such as advertising and marketing, but I hope you got the gist from these specific examples.
Potential vertical target markets
Even if all markets involve humans and therefore need intelligence, automation, and payment, some markets seem ripe for the fusion.
The first market that comes to mind is health care. Blockchain improves patients’ data security, while AI improves the analysis of, say, patients’ blood work or X-ray scans by detecting abnormalities early. Personalized medicine becomes possible by training AI algorithms on large amounts of data to be able to accurately analyze samples, taking into account a person’s genome. AI-assisted remote robotic surgery requires, for obvious reasons, the absence of single point of failures in the data network.
Another often cited market is supply chain management which currently suffers from a huge amount of paperwork, siloed data and therefore lack of analytical visibility. Despite some setbacks (slow adoption) with Maersk (project TradeLens with IBM) and Walgreens, the Fusion seems unavoidable since it improves connectivity and automation to securely track, record, and organize the movement of every ship and product in real-time.
Nothing else than AI can automatically process and deal with complex regulations from different countries, in various languages, while optimizing the delays, costs, and deliveries, taking into account, and predicting, customer demand, availability of materials. Blockchain, on the other hand, provides data security and visibility among all parties, proofs of ownership for expensive goods, both increasing trust, and collaboration. Such a complex global market also demands a solution without a single point of failure.
On a smaller scale, transportation faces some of those issues, and need scalable, open and efficient tracking of day-to-day deliveries using Blockchain, while AI provides accurate information about delivery times, taking into account (or predicting) traffic, weather, roadblocks. This technology is already used, at least partially, by many travellers, especially in self-driving vehicles.
Finally, Finance / Banking will benefit from safe and effective transactions. DeFi (decentralized finance) addresses current problems of traditional finance and provide better, faster and cheaper solutions. On the other side of the spectrum, the (infamous) CBDC’s will eliminate costly intermediaries (commercial banks) and improve speed thanks to directly interacting with central banks. “Cross-border transactions will become cheaper and faster, which will make the financial system more accessible to the unbanked”. (Let’s hope they keep their promise on that one…)
Deep Fusion
Ultimately, AI and Blockchain will merge totally, and achieve the full potential of deep Fusion, i.e. futuristic applications that cannot exist without both AI and Blockchain.
Singularity.net’s HyperCycle is a Cardano Layer2 chain designed to enable high-speed on-chain execution of optimized AI microservices. It will have profound implications for many domains beyond AI that require efficient and scalable sidechain solutions (i.e. solutions that do not run on the main blockchain, but ‘on the side’, for speed and scalability reasons).
Their AGI project “OpenCOG Hyperon” on Cardano will be a decentralized platform taking full advantage of an evolving blockchain infrastructure “where different AI strategies and methods such as Neuro Symbolic AI, Evolutionary Learning systems, Machine Learning, and other AI approaches can collaborate based on a shared knowledge graph”.
BITtensor is an open-source protocol that turns machine learning into a commodity “by leveraging the possibilities of a distributed ledger, specifically, its facilitation of open access/ownership, decentralized governance, and the ability to harness globally-distributed resources of computing power and innovation within an incentivized framework. Machine learning models train collaboratively and are rewarded according to the informational value they offer the collective.
Their ultimate vision is to create a pure incentivized market for AI, in which consumers and producers of this valuable commodity can interact in a trustless, open and transparent context”. In short, an “open-source repository of machine intelligence, accessible to anyone, anywhere, thus creating the conditions for open and permissionless innovation on a global internet scale”.
IBM is experimenting with turning smart contracts into “cognitive smart contracts” that can learn and adapt based on insights gained from AI, which identifies pattern changes in the data, recognizes meaningful interactions, detects suspect activities, etc. Clearly, such technology can improve the effectiveness and value of DAO’s.
Matrix.io goes even further by “merging blockchain and AI into a single application”. Such technology will “make blockchains smarter and add the ability to evolve through self-learning”. Their service platform provides AI algorithms, or an AI-assisted NFT generating and trading platform.
They are even working on a “general-purpose platform that allows users to upload their brain waves data to create their individual avatars” (like in the movie “The matrix”) which would give the Metaverse an even more profound impact on humanity.
You’ll find more information about the Fusion, and applications in the already mentioned paper ‘Blockchain: The Next Breakthrough in the Rapid Progress of AI’.
Conclusion
This research should be an eye-opener for anyone. It is clear that AI and blockchain will fuse more and more, and affect every aspect of our lives (let alone when they fuse with IoT), and thereby change society profoundly, in ways difficult to predict.
The Fusion will foster the emergence of disrupting applications, and the speed of such changes suggests that startups will emerge as winners against Big Tech mastodons, at least in the most innovative aspects of this revolution.
Not only companies are developing solutions to specifically target their current monopoly over people’s private data and computation power. Flexibility is an asset,as we recently saw with ChatGPT, created by OpenAI, a very small company compared to Google, that nonetheless managed to take the lead in AI, and is now being spun into pretty much any application under the sun.
I recommend reading this excellent article ‘We have no moat’, in which Google admits that neither they, nor OpenAI for the matter, will win the race. I’ll leave you ponder what they stated:
The uncomfortable truth is, we aren’t positioned to win this arms race and neither is OpenAI. While we’ve been squabbling, a third faction has been quietly eating our lunch.
I’m talking, of course, about open source. Plainly put, they are lapping us. Things we consider “major open problems” are solved and in people’s hands today:
LLMs on a Phone: People are running foundation models on a Pixel 6 at 5 tokens / sec.
Scalable Personal AI: You can finetune a personalized AI on your laptop in an evening.
Responsible Release: This one isn’t “solved” so much as “obviated”. There are entire websites full of art models with no restrictions whatsoever, and text is not far behind.
Multimodality: The current multimodal ScienceQA SOTA was trained in an hour.
While our models still hold a slight edge in terms of quality, the gap is closing astonishingly quickly. Open-source models are faster, more customizable, more private, and pound-for-pound more capable. They are doing things with $100 and 13B params what we struggle with at $10M and 540B. And they are doing so in weeks, not months.
This is all about AI, sure, but now combine it with the key advantages that only Blockchain brings to the table, and it’s easy to imagine the breakneck speed at which the world is going to evolve. It’s the perfect storm for which I’ve been preparing for the last 3 years…
Which brings me back to my Omdena Challenge, that I started because not only because AI is already complex and not well known by many developers, but Blockchain is a totally different world, full of new technologies, complex protocols, encryption algorithms and decentralized applications, to the point a new term had to be created for them, Web3.
In my opinion, we are going to witness Web3AI going nuclear!
Talking about nuclear… looks like Diffusion models agree, kinda.
I hope you enjoyed this (long) article, I certainly did.