We use cookies on all our websites to gather anonymous data to improve your experience of our websites and serve relevant ads that may be of interest to you. Please refer to the cookies policy to find out more.

By continuing, scrolling the page or clicking a link, you agree to the use of cookies.

Artificial Intelligence and data capture

10 July, 2018Ouida Taaffe

Male-working-on-his-laptopWhat is the best thing since electricity? According to Manoj Saxena, chairman of AI start-up CognitiveScale (and, before that, the first general manager of IBM Watson), it is artificial intelligence. “Artificial intelligence is one of those fundamental shifts in technology. It will make all data management processes more efficient…it will go all the way to the back office and to core systems,” says Saxena.

AI vs traditional data capture

Saxena points to three main ways in which AI differs to existing data capture. First, AI can tackle both structured and unstructured data. Traditional systems chew through text and numbers, whereas AI can extract useful information from, say, pictures and recordings of speech. Second, AI can deal with both probabilistic and deterministic information – much as human beings deal with both guesses and with facts. Third, where traditional data systems are based on rules, artificial intelligence is based on patterns, which means it can evolve. The way this is done is to ‘train’ the system.

Training a system to recognise a particular pattern takes a lot of relevant data and a lot of computing power, which is partly why AI is really only being put into widespread practice now. Fifteen years ago, a system that selected tabby cats and ignored tigers would have been both expensive and difficult to programme. Now, it is relatively trivial. It is possible for machines to ‘learn’ what a picture of a cat looks like without anyone specifically labelling the pictures.

Finance Services

But do systems that can sort cat pictures have any relevance to financial services? “AI is going to transform Wall Street and financial markets by orders of magnitude,” says Saxena. The question, however, is whether banks will be able to make use of all the capabilities of AI. They certainly have data and computing power and the ability to invest, but there are not that many people in the world who can build advanced AI systems – around ten thousand, according to Saxena.

That is where CognitiveScale wants to come in. The aim of the company, which has IBM, Intel and Microsoft among its investors, is to be akin to an app store for enterprise AI systems. The idea is that AI ‘building blocks’ will enable someone with an undergraduate degree in computer science to build advanced AI systems. “That democratisation of skill also happened in the internet,” says Saxena. “At first building a web page was complex and time-consuming, now one can be published in less than a minute.”

AI Future

AI promises to bring many changes – among them hyper-personalisation of consumer products and improvements in back-office processes thanks, in part, to the ability to predict consumer demand. The mandatory introduction of APIs – via Open Banking and PSD2 – will allow similar product tailoring in financial services and could bring profound changes.

“I think financial services will start morphing into wellness and entertainment. You are already seeing that in China,” says Saxena.

Competition to provide that sort of deep personalisation at speed and scale will also change bank systems. To feed APIs (both coming and going), core banking systems will have to be able to store more data and different forms of data. That does not, however, mean that banks will swap out their current systems wholesale. “Core banking systems may not be the most advanced available, but they have a lot of know-how on regulation and security, on what does and doesn’t work, and they are battle-tested,” says Saxena. He expects microservices and APIs to be provided by vendors on a virtualisation layer and changes to the core systems to come over time.