Pauline E. Ranger AI is familiar to us as consumers: we enjoy cars that can parallel park themselves, devices that respond to questions, and streaming platforms that suggest shows that we might like. Professionally, however, the bigger question is how industries will harness the power of AI. AI in finance offers operational efficiencies for everything from risk management to trading and insurance underwriting.
Artificial Intelligence – What is AI?
Artificial intelligence is a field of computer science that focuses on creating intelligent machines that function like humans. AI computers are designed to perform human-like functions, such as Learning, decision-making, planning, and speech recognition.
Artificial intelligence allows machines to continuously improve their performance without the need for humans to provide prescriptive instructions on how to do this. This is important for two reasons. Humans know more than they can tell. Humans may be able to recognize a person or use a clever strategy when playing chess. Prior to the development of advanced artificial intelligence, we were unable to automate many tasks due to our inability to express ourselves. Second, AI technology has superhuman performance, often operating faster and with greater accuracy than humans.
Artificial Intelligence Technologies
Artificial intelligence is a broad term that encompasses many capabilities and technologies. PWC, a consulting firm, confirms its claim that AI is not a “monolithic topic area.” AI is a collection of different things that contribute to the idea of ‘intelligence.’
Machine Learning is an automated method for building analytical models. Using algorithms that iteratively learn from data, [machine learning](https://www.toptal.com/machine-learning/machine-learning-theory-an-introductory-primer) enables computers to find hidden insights without being explicitly programmed where to look.
Deep Learning is a subset of Machine Learning. It has improved object recognition, video labeling, and activity recognition. It is also making progress with perception (audio and speech). Facebook’s DeepFace, for example, has been -trained to recognize faces in photos. Deep Learning is often compared to biology. However, experts agree that, while inspired by the brain, it is not the same.
Natural Language Processing refers to the ability of computer programs to comprehend human speech in real-time. Research and development is moving to systems that can interact with people via dialog rather than just responding to stylized requests.
The Internet of Things (IoT). It is dedicated to the idea of interconnecting a variety of devices, including appliances, cars, and buildings. If your alarm goes off at 7:00 am, your coffee maker could be notified to begin brewing your coffee. It is also a part of the larger trend. Wearable technologies act as sensors.
This list is by no means exhaustive. Below, you will find a variety of topics and technologies related to artificial intelligence.
Artificial Intelligence Market size
In the Harvard Business Review article, it is predicted that “the effects of AI in the next decade will be magnified as manufacturing, retail, transportation, finance and health care, law and advertising, insurance and entertainment, education and virtually all other industries transform their core business processes to take advantage machine learning.” The bottleneck lies in management, implementation, and business imagination.
It is expected that the widespread adoption of AI in all industries will drive global revenues to $12.5 billion by 2017 and $47 billion by 2020, with a CAGR of 55.1% between 2016 and 2020. The industries investing the most will be banking and retail, followed by manufacturing and healthcare. These four industries, combined, will account for over half of the global AI revenue in 2016. The banking and retail sectors alone are expected to generate nearly $1.5 billion each.
The greatest AI investments across industries will be made in areas like automated customer service agents (ACSAs), automated threat intelligence, and fraud analysis. According to Jessica Goepfert, program director of market-research firm IDC, “Near-term opportunities for cognitive systems exist in industries like banking, securities, investments, and manufacturing.” These segments have a lot of unstructured information, as well as a willingness to embrace innovative technologies.
Artificial Intelligence and Finance: Present and Future Applications
Artificial intelligence could be used to improve operational efficiency in finance, from trading and risk management to claims and underwriting. Some applications are better suited to certain sectors of financial services, while others can be used across the board.
Artificial Intelligence and Finance: Risk Management
Artificial intelligence is a valuable tool for security and fraud detection. Computers are used to analyze structured data and compare it with a set of rules. A payments company may set a threshold of $15,000 for wire transfers so that transactions exceeding this amount are flagged for investigation. This type of analysis is not without its flaws, and it requires additional work. Even more importantly, cybercrime scammers often change their tactics. The most efficient systems are those that continually improve.
Using advanced algorithms such as deep Learning, it is possible to add new features for dynamic adjustments. Samir Hans is an advisory principal with Deloitte Transactions and Business Analytics LLP. He says that cognitive analytics can make fraud detection models more accurate and robust. When a cognitive system flags something as fraud, but aa human judgeit not to be fraud for X, Y, or Z, then the computer will learn from that human insight and won’t make a similar determination next time. “The computer is becoming smarter and more intelligent.”
PAYPAL’S SUCCESS IN ARTIFICIAL INTELLIGENCE AND FRAUD DETECTION
PayPal, the payment giant, has advanced fraud protocols. PayPal’s size and visibility make it a target for fraudsters. In 2015, PayPal processed $235 billion from 4 million transactions made by its 170,000,000 customers. PayPal was able to increase security through deep learning technology. PayPal’s fraud rate is 0.32 percent of revenue. This figure is much lower than the average merchant experience, which is 1.32%.
PayPal has used linear, simple models in the past. PayPal’s algorithms today mine data from the purchase history of a customer and review patterns of fraud that are stored in its ever-growing databases. A linear model may only require 20-30 different variables, whereas deep Learning can handle thousands of data points. These capabilities allow PayPal to distinguish between innocent and suspect transactions. Hui Wang is PayPal’s Senior Director for Global Risk Sciences. “What we like about more advanced machine learning, it’s ability to consume more data and handle layers and levels of abstraction, and to be able’see’ […] that even humans might not be capable of seeing.”
TRANSITION FROM HUMAN CONSTRUCTED MODELS to TRUE AI
Investment management companies have been using computers to execute trades for years. Around 1,360 hedge funds represent 9% of funds and rely on large mathematical models developed by data scientists, often with mathematics PhDs. These models are static and only use historical data. They also require human input, and they don’t perform well when markets change. Therefore, funds are increasingly moving towards artificial intelligence models that can analyze large volumes of data and improve themselves.
These new technologies use complex techniques, including deep Learning and Bayesian Networks. They are also inspired by genetics. AI trading software is able to absorb huge volumes of data in order to make predictions and learn about the market. To better understand global trends, they can use everything from books to tweets, news, financial data, and earnings numbers. They can even watch Saturday Night Live sketches.
It is important to note that the above is not high-frequency trade, which allows traders to execute millions of orders, scan multiple markets, and respond to opportunities in a way humans cannot. AI-driven platforms are looking for the best long-term trades, and machines, not humans, dictate the strategy.
Startups have developed some of these AI systems. Hong Kong’s Aidiya, for example, is a hedge fund that uses artificial intelligence to make all its stock trades. Ben Goertzel is the co-founder of AI Trading. Goldman Sachs was the lead investor in Series A and installed an AI trading platform named Kensho. Kensho’s Series B round included Wall Street’s six largest banks, including Goldman Sachs. JPMorgan Chase and Bank of America Merrill Lynch. Morgan Stanley, Citigroup, and Wells Fargo also participated.
