How Is AI Being Used In Crypto?
What is AI?
Artificial intelligence (AI) is a branch of computer science that builds machines capable of intelligent behaviour, replicating or simulating human intelligence. AI has been around since the 1950s, but almost 60 years on it is now smarter than ever. Not only do we have AI, but data driving “machine learning”. This can be used in a lot of different fields. Below we try to summarize the aspects, benefits and risks of AI and its uses in crypto
In 1950 Mathematician Alan Turing posed the question: “Can machines think?”. Later Turin’s paper and subsequent tests layed out the goals and vision for artificial intelligence.
An intelligent computer uses AI to think like humans and perform human tasks, whereas machine learning is how a computer develops its own intelligence. Stanford University defines Machine Learning as “the science of getting computers to act without being explicitly programmed” to do so.
Bernard Marr (author and strategic business and technology adviser) claims that the potential in AI is far greater than many of us believe or know yet. “It allows computers to see, it allows computers to hear, it allows machines to walk”. Now AI and machine learning have begun to merge together and create powerful digital tools that can be used in crypto and beyond.
How does AI work?
AI consists of an amalgamation of different principles. The key components that link all of these tools together though are algorithms. These computer instructions are the foundations to any computing system and also play an important role in supporting AI systems and functions. When we are talking about inputting data into AI this is turning human intelligence into computer code or algorithm that digitally represents that information.
Machine learning is a sub-domain of AI who’s application enables systems to automatically learn and improve from experience without specific input data like seen in ANIs. Machine learning uses direct experience and instruction to then create and learn patterns in new data in order to “self-learn” without constant human intervention.
Examples of this process are platforms like Google, Netflix and Spotify that all learn from your searches and generate suggestions and find ongoing patterns in you as a user.
Deep learning is a sub-domain of machine learning and is related to algorithms called Artificial neural networks (ANNs). This is a circuit of nodes or artificial neurons directly that was first inspired by digitising biological brain systems. In ANNs there are a series of algorithms that process data between different underlying variables and process the data like a human brain would. Deep learning has been applied to assist in voice recognition, computer vision, social network filtering and more.
Natural language Processing (NLP) is a science of interpreting, understanding and reading a language using a device. Once a machine understands how users might intend on communicating it can respond accordingly.
Computer Vision relates to the systems that get information from images, videos or multi-dimensional data (3D). Computers here can gain understanding of digital visuals by breaking down the objects and making observations based on previous input. Face recognition could be an example of this.
Cognitive Computing algorithms animate text, speech, images or objects in a way that a human brain would. This then tries to give an output aligned with an expected outcome.
How can AI be used in crypto?
We know market volatility can worry a lot of people, especially if you are new to crypto. AI has been a powerful tool emerging in the finance industry and seems to be even more so when combined with blockchain. This is because the blockchain keeps a permanent digital record of every transaction which makes data accessible to analysis with AI.
We can use AI and machine learning to better understand market predictions and make them more accurate. The key is in the data collection and analysis. The more history there is, the more accurate the predictions can be.
E.g. Crypterium predicts 150+ digital currency prices powered by AI. They use a three-fold system that collects, analyses and then predicts market prices with a self-learning algorithm. It is a subscription based tool with different plans available.
AI can help optimise crypto transactions. Using wide network analysis it could better determine the best time of the day to perform a transaction to save on gas fees. This assists a combination of price and time optimisation for faster and more efficient transactions in a protocol. The key to AI success is: Will the AI tool do the job faster and more efficiently than a human can?
E.g. CryptoHawk is also an AI price prediction platform but has some key features that help users optimise their transactions. The technology allows investors to navigate trends, market volatility and gain competitive advantages. CryptoHawk have tools specifically for BTC and ETH, as well as for Altcoins and TradeWatch for strategy and optimisation.
The collection and analysis of vast amounts of data across blockchain networks also leads to the general publication of useful statistics. Using AI tools we can collect and sort the relevant data to create statistics through automated surveys etc.
Analysing user behaviour
Alongside studying general market trends and big data across whole networks, closely analysing user behaviour is something that AI can do; this is also linked to machine learning. Once AI has been implemented with a set of relevant algorithms that show certain behaviours, machine learning technology keeps the ball rolling, so to say. It uses the information to find similarities, differences and identify trends in behaviour. This digital skillset is the fulcrum to understanding user behaviours in an unbiased way.
Automation AI is already quite widely used. It can be used in customer care; you know when you see a chat box pop-up when visiting a website? This is a chatbot. They are automated systems that respond to common problems and respond with programmed answers and responses. For companies this is a very useful way of filtering customer care problems to relevant help services. It could also be set up to automate transactions (linking to the time and price optimization) so that users can buy, sell or trade when they are not directly online.
E.g. Crypto Hopper is the leading automated trading bot in the market that hosts more than 9M transactions per month. Crypto trading happens 24/7 which can be stressful for traders wanting to have the best opportunities throughout a day. Crypto Hopper uses unique AI to recognise the trends and monitor your assets accordingly.
Possible risks of AI
One of the “problems” with us humans is we are liable to make biassed decisions. It's human nature to have an opinion and act accordingly to this. When we are assigning responsibility to AI however, it is really important to address these biases. This is particularly relevant to crypto as by nature users won’t want to use a network favouring those in the US for example.
Loss of jobs
When we replace jobs with computer generated automation there are concerns that there will be an ongoing job loss. Realistically it will also create space for new jobs too.
Studies by both PwC and Mckinsey & Co. both found that the amount of new jobs that will arise will actually exceed the amount lost. It is true that the new jobs will be different and require different skill sets in training.
It's useful to think of it as a reshuffling of responsibility - the job market is an ever morphing entity whose needs change over time. Introducing AI technology into everyday use will change the distribution of jobs that need to be “human”.
Hacking is always a huge concern in crypto. While sometimes it can be used positively to “check” the reliability of a new product for example, we would want to make sure that when we increase AI use, safety will not be compromised.
AI and machine learning could also be used to take advantage of networks in cyber attacks, as AI’s can learn and analyse behaviour. This can mimic and automate an attack or create a bias that will compromise users or whole systems.
On the other hand though, AI systems could be set up to detect this kind of activity and alert companies when this behaviour is being seen. This way perhaps cyber attacks could even be reduced.
The Future of AI
There are many different types of AI machines used today, and the potential in the future can’t yet be determined. There are also several studies and attempts to use AI in crypto and in particular in the DeFi ecosystem to improve impact, reliability, and security.
But one question many are keen to answer is whether, or perhaps when, AI will be able to move past merely thinking for itself and become self aware. Check out ALLTIME10S video for a short insight into future possibilities for AI.
29 March 2022