Artificial intelligence (AI) has the potential to radically change the financial world. Companies spend billions of dollars on a technology that they hope will help them make smarter decisions, improve customer service, reduce fraud, and - they hope - save a lot of money.
The company Autonomous estimates that the financial industry will be able to cut its costs through AI by 22 percent by 2030. That's about a trillion dollars.
However, the financial sector is at a very early stage in this process and is just beginning to incorporate technology into its systems. Large financial institutions such as Bank of America or Capital One use AIs for chatbots or voice assistants to answer customers' simple questions about their money, such as money transfers. Paypal uses similar technologies to give its customers a more detailed insight into their transactions. Paypal saves 25 million dollars a year.
Companies also use the technology to make informed decisions about who to lend money to - and who not to. Banks are working flat out to organize and clean up their data and promote the cloud technology needed to analyze huge data sets. The bank executives hope that the machines will be able to evaluate enormous market data sets and Bloomberg chats to present customers with tailored investment ideas.
JPMorgan does not want to afford missteps in artificial intelligence and machine learning
Samik Chandarana, who has been head of data analysis, artificial intelligence, and machine learning at JPMorgan's Investment Bank since 2017, stresses that machine learning will not answer every open question. But it's also an important lesson when you look at the increasingly high expectations of New Tech. Wall Street saved 41.1 billion dollars in artificial intelligence in 2018. This is the result of a report by IHS Markit. The global market value of AI will rise to $300 billion by 2030.
So how should a company apply new technology?
Chandarana and his colleagues, Magu, Head of the Center of Machine Learning Excellence at JPMorgan, and Veloso, Head of AI Research at the Bank have developed a three-step process to evaluate, test and integrate techniques from the world of artificial intelligence into his company. With a tech budget of $11.4 billion in 2019, JPMorgan has the largest war chest to invest in artificial intelligence and machine learning. But this also means that the bank has a lot of money to lose if it does not develop the technologies wisely.
Chandarana overlooks the first stage, bringing data specialists together with colleagues from headquarters and field offices who are in direct contact with clients. The data specialists work with teams from the market, payment or application segments to understand what their colleagues are doing there and how Artificial Intelligence can help them. Chandarana has avoided building parallel teams to eliminate problems that should be better tackled together.
Mangu, who spent 17 years at IBM's Watson Research Center before joining JPMorgan, overlooks the second step. The latter relies above all on technical experts who come from different areas of artificial intelligence. The team supports the group from the first step with language analysis, language-to-text formats and deep learning by translating documents and research into usable computer code that can be used in all departments of the bank.
The last step is particularly theoretical. It is conducted by a research team headed by Veloso, an elite academic hired by the bank in 2018. Your team is trying to explain how an algorithm comes up with a solution. They thus explain the algorithm's ability to interpret and lead the discussions about ethics and fairness. The team also helps with the first two steps when in-depth research and analysis is needed.
It is important to clarify where new technology should not be used.
The goal of this strategy is to successfully implement AI technologies. But it is equally important to identify the areas where the technology should not be used. JPMorgan has created logging systems for each project.
Whether data is not available in a usable form or a project is not worth following up - all these circumstances must be recorded in writing so that mistakes are not repeated.
This article was posted on Business Insider.