Intelligent Cryptocurrency
Using Intelligent Cryptography
One of the catchphrases of the software industry's past ten years is "Software is devouring the world."
It summarized the notion that businesses that primarily functioned in the physical world were shifting to the digital economy in a trend that would effectively convert every firm into a software company. It was cited in 2011 by software legend and venture capitalist Marc Andreessen.
CEO of the blockchain and cryptocurrency industry research company IntoTheBlock, Jesus Rodriguez. The discussion he will give this week on the Big Ideas stage at Consensus 2022 in Austin, Texas, is previewed in this article.
Recent years have seen a rapid advancement in machine learning (ML) and artificial intelligence (AI), which has led several experts to assert that "machine learning is devouring software."
Since cryptocurrencies and digital assets are built on a foundation of code and programmability, ML-AI developments are expected to have an impact on them.
A new age in which intelligence becomes a fundamental part of crypto assets is anticipated, to begin with, the convergence of ML-AI and digital assets.
Though conceptually simple, the concept of intelligent crypto assets is fraught with real-world difficulties. What are some of the basic ML developments that have the potential to quickly influence the upcoming wave of crypto assets?
What are some of the primary technical obstacles that need to be removed for crypto to become intelligent, or what are the main scenarios that may benefit from crypto's intelligence capabilities?
This essay investigates some of these concepts and offers a thesis regarding the promise of the crypto-machine learning (ML) nexus.
Only cryptocurrencies have native intelligence.
Understanding that crypto is the first asset class in history with the potential to become natively intelligent is crucial when thinking about AI-ML in the context of digital currency.
Traditional asset classes, like commodities or equities, use AI-ML capabilities that are deployed in tools like Robo-advisors or quant strategies that exist outside of the actual asset.
Even if these vehicles clearly have a place in the crypto arena, crypto-assets can already come with built-in AI-ML capabilities.
This gain is undoubtedly a byproduct of cryptography's programmable and digital capabilities. The foundation of cryptocurrency assets is code, and that code may take the shape of AI-ML models.
Crypto will be eaten by machine learning, but how?
In the coming decade, the crypto sector is projected to experience significant growth because of AI-ML. The earlier stages of cryptocurrency have mostly focused on automation and digitalization, but the following iteration is destined to be intelligent.
Although there are many AI-ML applications in the cryptosphere today, we cannot assert that cryptographic assets have built-in intelligence.
We should prepare for the implementation of AI-ML as inherent capabilities in crypto-assets and protocols in the near future, enabling them to learn from their surroundings and adjust their behavior in response to markets or other factors.
The astounding development of AI-ML technologies over the past several years has partly determined the inevitable development of intelligent digital assets.
We should consider AI-ML in the context of cryptography as a collection of interconnected sorts of procedures rather than as a single, general concept. Accordingly, there are only a few AI-ML universities that seem especially well-suited for use in the crypto industry.
Let's examine some of the most well-liked methods within the context of their possibilities in cryptographic technology.
Transformers
Transformers, widely regarded as the most significant advancement in AI-ML over the past ten years, are responsible for the revolution in natural language understanding (NLU) and are also making progress in other fields, such as computer vision.
Models like OpenAI's GPT-3 or Nvidia's Megatron can produce artificial writings that are indistinguishable from those produced by people, interact with questions and answers in extremely complicated ways, and even demonstrate reasoning ability over textual forms.
Models that can create creative pictures from textual forms include Google's Imagen and OpenAI's DALL-E 2, which connect intelligence across several disciplines.
It's not difficult to foresee the influence transformers will likely have in fields like NFTs that rely on visual representations and textual interactions given their importance in the NLU and computer vision arena.
Self-directed instruction
Self-supervised learning (SSL) was recently referred to as the "black matter of AI" by Meta (Facebook) AI Research as an analogy concerning the fundamental function that this novel sort of approach can play in the upcoming generation of AI models.
SSL's conceptual goal is to provide users access to intelligence in a way that is similar to how infants learn via observation and interaction.
Traditional supervised learning techniques require a significant amount of labeled data to be trained, which is one of the drawbacks of SSL.
Without prior training, models like Meta's DINO can identify objects in photos.
Crypto seems to be a wonderful application for learning without a ton of labeled data. These techniques may immediately benefit decentralized finance (Defi).
Networks of Neural Graphs
Blockchain datasets are the primary source of data for the cryptocurrency industry. Blockchain datasets are naturally hierarchical from a structural perspective since they represent connections between addresses, transactions, or blocks.
The branch of AI-ML that specializes in learning across hierarchical datasets is called graph neural networks (CNN). guns are being used by businesses like Google's DeepMind to anticipate traffic on Google Maps or even to comprehend the structure of glass.
GNNs appear to be the ideal AI-ML method for crypto assets. guns are probably going to be crucial in helping blockchains learn from their native datasets if they ever want to be intelligent.
Reinforcement in education
Following Lee Sedol's defeat by DeepMind's AlphaGo, deep reinforcement learning (DRL) gained some notoriety. AlphaGo learned to Go by defeating itself in an inconceivably enormous number of games and fixing its own errors. The core of DRL is this iterative, interaction-based learning process.
Many aspects of cryptography, such as Defi or NFTs, where circumstances are constantly changing, seem to be compatible with the DRL concepts of learning via mistakes. Because DRL is so good at games and because the majority of crypto algorithms are founded on sound game-theoretic principles.
Science fiction author William Gibson, a pioneer of cyberpunk, famously observed, "The future is already here; it's just not fairly distributed.
" That passage may act as a philosophical road map for us as we consider the development of intelligent crypto assets. Cryptography was developed during the height of AI-ML research and technological advancement.
Today's AI-ML technologies are quickly gaining popularity, and it won't be long until they are treated equally to other crypto technologies. There appear to be use cases everywhere. Let's talk about some of the most evident.
Advanced NFTs
Some applications of leveraging AI-ML generative techniques to produce non-fungible tokens have been made (NFTs). But the NFT space as a whole should feel the impact of AI-ML.
Imagine NFTs with language and speech skills that can engage in conversation with people, respond to inquiries about their meaning, or interact with a particular setting.
Imagine conversing with a visual NFT that can alter its look depending on the nature of the interaction, much like you would with your favorite digital assistant.
Consider employing AI-ML transformer models that have been trained on millions of paintings to create distinctive NFTs that perfectly replicate the distinctive features of the artists' styles.
Defi intelligent protocols
Although the protocols for decentralized finance (Defi) are all about automation, they are not particularly sophisticated. It seems likely that Defi protocols will incorporate AI-ML capabilities.
A new generation of automated market maker (AMM) protocols that can modify pool balances based on real-time prediction models and current market circumstances is something we can imagine.
Similar lending methods come to mind, which modify loan amounts based on an in-depth analysis of the addresses making the request.
L1-L2 intelligent blockchains
Blockchains are not going to be an exception to how AI-ML is affecting many facets of software infrastructure, including networking, computation, and storage.
The idea of intelligent consensus procedures that enhance performance based on predictive models is not implausible. Similar blockchains that create intelligent economies to manage the cost of computing in the form of "gas" or other equivalencies come to mind.
Crypto-savvy applications and apps
One of the most apparent places to use AI-ML skills seems to be user experience. Before wallets or exchanges begin integrating native intelligence capabilities that aid in improving investing and trading decisions that are currently wholly dependent on human subjectivity, it is only a matter of time.
Programmable, intelligent stablecoins
Following the failure of the Terra UST, programmable stablecoins appear to be a hot issue right now. What if, instead of seeing this stablecoin type as programmable, we might consider forms that are both intelligent and programmable?
What if stablecoins could rely on AI-ML algorithms that naturally learn from market situations rather than programmed stablecoins that alter the peg according to statically set economic gymnastics?
An intriguing strategy to investigate in this field seems to be the mix of AI-ML with human supervision.
Contrary to popular belief, there is a more reciprocal link between cryptography and AI/ML.
Although there are some very apparent ways that AI-ML might affect the future generation of crypto assets and infrastructure, there are also less obvious ways that crypto can impact AI-ML.
Decentralized AI is an emerging technological trend that aims to use tokenization and decentralized computing to address some of the growing centralization issues with AI-ML technologies.
Mechanisms that use crypto assets to build ecosystems where businesses and individuals are rewarded for sharing data and AI-ML models are a subdomain of the overall decentralized AI approach.
Data powers AI and ML, but it is tightly regulated by a limited number of established players, and there are almost no incentives for businesses to cooperate and exchange data to end this monopolistic cycle.
Smart tokenomics and incentive systems might naturally open up channels for businesses to collaborate on the development and training of AI-ML models for certain tasks and split the rewards.
Another current hot subject in AI-ML is bias and fairness, which might be greatly impacted by the adoption of native crypto technology.
Datasets used to train AI-ML models are rife with prejudices, discrimination, and poisonous data points that might affect the accuracy of the models' knowledge.
Even though there have been significant improvements in assessing and monitoring the fairness of AI-ML models, the industry as a whole lacks reliable benchmarking and accountability systems.
Imagine monitoring the bias and fairness scores of certain AI-ML models using a blockchain layer, and rewarding models that are increasing their fairness ratings. For the use of blockchain technology in AI-ML infrastructures, this is a low-barrier situation.
The future generation of digital asset technologies should unquestionably include AI-ML as a core component, but there is also a lot of practical value that blockchains and cryptocurrencies can offer in the field of AI-ML.
Fundamentally, crypto might act as a financial and accounting framework for developing AI-ML solutions that are more equitable and democratic.
Every aspect of the software industry is being impacted by AI-ML, and crypto is not going to be an exception.
The fundamental ideas of digital asset technologies have been focused on employing automation and digitalization to make financial services more accessible to everyone.
One of the crypto industry's newest frontiers is intelligence, and its effects are likely to be seen throughout the industry as a whole.
The use of AI-ML is anticipated to spark a new wave of innovation in the crypto industry, including intelligent NFTs, Defi protocols, and new types of crypto assets.
There are already available technologies and application cases. It's time to begin construction.
Comments
Post a Comment