Katharine Wooller, Chief Strategist, Financial Services, Softcat plc discusses how Blockchain underpins AI evolution in Financial Services
The intersection of blockchain and AI is truly fascinating; the two emerging technologies are increasingly intertwined, and it is likely that decentralised technology is the key to unlocking the huge potential of AI.
For those defi long-timers, like me, the criticisms of AI do sound rather familiar! It is not news for a radically different and socially disruptive technology to have its detractors – dramatic headlines pour over the need for regulation and the potential for harm. However, as the two technologies become increasingly established it seems to me that they need each other, and that they are hugely powerful when used jointly. In short, blockchain’s transparency, security, and decentralisation, fixes what’s wrong with AI.
Huge market and potential upside
The two technologies are synergistic, and the potential market is huge: a UN Trade and Development report estimates that the AI market will be is projected to hit $4.8 trillion by 2033 and will be the dominant frontier technology. Crucially, it estimates that AI will account for 29% market share, whilst blockchain technologies will have a 14% market share. Interestingly, that suggests significant potential growth in AI related coins, such as Near, Artificial Superintelligence Alliance, and Filecoin, which at the time of writing have a $30bn market cap (source Coingecko).
A perfect marriage
From a technical point of view AI and blockchain work together harmoniously. Blockchain provides immutable ledger capabilities to keep the AI’s predictive and analytical capabilities honest. This is critical for a number of reasons. From a data integrity point of view, blockchain can improve the data that AI consumes, by ensuring data provenance and integrity which helps avoid the “rubbish in, rubbish out” problem in machine learning models. In my day job, supporting innovation and transformation at over 2,000 financial services businesses, the first question for any business looking to introduce AI should be “is our data AI ready?” – the answer, more often than not, is “not yet”!
Problem Statement
AI comes with some unique problems that will require novel solutions. The core of the issue is that data, and how AI uses it, is rife with ethical issues.
Most importantly, AI is only a good as the data it is trained on and is open to bias. Depending on the information they are given they may be sexist, ageist or homophobic. If you are using AI to work out who to give a mortgage to, or allow to vote, or for recruitment, that’s a huge moral conundrum. AI can be intentionally poisoned for nefarious reasons or simply hallucinate.
Secondly, who owns that data, and whether they can be trusted not to abuse that power and how can prevent harm are key concerns. The foundation stone of blockchain is decentralising data, knowledge, and power – concepts that translate neatly to the data needed for AI.
Crucially, the current big AI models are controlled by a few large players. Microsoft, Google, Amazon, Meta (formally called Facebook, which is thought to have 52,000 data points, on average, for each user). Rightly there are concerns that this concentrates power, and that the manipulation of data being used for an AI model can be political -perhaps by using organised misinformation to swing an election which is already a close call. Indeed, Facebook has been accused of interfering in US elections just shy of 40 times. The question is, do we trust big companies and governments to always do the right thing? Decentralisation and blockchain, keeps the decision-making honest, particularly in a “blackbox” scenario.
Thirdly, how do reward those whose data is used to train models? The reason those big companies control AI models is that they have the data to train models and can afford the computational power to do so. There has been much handwringing over how AI models are trained, specifically who owns the data, how do ensure privacy, and how are the owners of that data rewarded. Blockchain can be a panacea for these moral quandaries.
Decentralised AI marketplaces can provide a platform where data providers can transact directly, increasing accessibility for data whilst maintaining control and transparency. By using blockchain based learning methods, the models can be trained on decentralised data without direct access to it, ensuring privacy. It provides a way of paying those who own the data, without giving away the keys to the castle. There are some very interesting projects looking at decentralising power consumption, using compute when it is spare, to keep the environmental cost to a minimum.
Fixing Industry specific pain points
Whilst privacy and data ownership should be prioritised regardless of business type, there are some verticals that suggest a perfect blockchain use case.
Financial services firms should be concerned about models being trained on highly sensitive and easily recognisable personal data; blockchain can anonymise this data. Interestingly AI, when used correctly, can be used to huge positive effect in financial services businesses, and AI can be used to enable finding fraud in real time, and automating compliance data.
Healthcare, similarly, has data can provide a goldmine of insights, particularly for training models for AI assisted diagnosis, however the data owners do not want deeply personal data (such as photos) exposed without controls. For any business with supply chain issues, be it rare foodstuffs, posh handbags or vaccines, the more quality data the better, and using blockchain will ensure large supplies of quality data can be accessed easily and ethically.
Managing risk and future proofing
However, deploying blockchain technology is not without its challenges and risks. A few years ago, I was lucky enough to address a room full of FTSE 250 CISOs, around business risk related to blockchain – at the time around 99 of the world’s top 100 businesses by market cap were reported as experimenting with blockchain – but very few of the audience were actively considering the unique associated risks of the technology.
Scalability and interoperability are still a concern, as is the complexity of integration. Anyone who has tried to hire for AI or blockchain skills recently will tell you that there is a dearth of specialist technical skills in the UK and Europe, not made any easier by the speed at which the technologies are developing. Regulation for both technologies is still developing, and there is understandably some reluctance to invest heavily in technologies where the compliance requirements are an ever-moveable feast.
Further, the short term and medium-term outlook for bleeding edge technology focuses massively on quantum computing. It poses a huge challenge for cybersecurity and all blockchain technologies, when in the next few years, it is likely that advances in quantum computing will break the encryption we rely on. Equally, it is likely that quantum computing will massively accelerate the adoption of AI.
Brighter Days
Most industries and businesses are already deploying AI and will continue to do so. AI and blockchain is a marriage made in heaven. AI brings intelligence and automation – it will derive huge efficiencies and ultimately the advancement of mankind. Blockchain keeps it honest bringing trust, transparency, and decentralization.
AI, with the correct checks and balances brought by blockchain, will deliver a fantastic future.
The two technologies need each other, and long may it last.
About Author:
Katharine Wooller is Chief Strategist, Financial Services at Softcat plc – A FTSE-listed IT company – www.softcat.com. Katharine is a renowned author and speaker on AI in Financial Services.
