Humans have long attached great value to intelligence, as the Latin representation of the human species as homo sapiens, or ‘wise man’, illustrates. This is reflected in a desire to understand the world and to train our actions so that we can be as effective as possible in achieving our objectives.
AI as a discipline goes a little further than that: we attempt not only to understand the world and train our own actions, but also to build intelligent entities (or ‘agents’).
The intelligent (or ‘rational’) agent is central to the concept of AI and its practical application. An agent is something that uses a sensor to perceive an environment and an actuator to act or respond. For a human agent, the sensors may be its eyes, ears, touch or taste, and the actuators may be its voice, hand or foot.
For a robotic agent, the sensor may be a camera or a range finder and the actuator may be a kinematic arm or a motor.
For a software agent, the sensor may be a keyboard, joystick or voice-recognition system and the actuator may be the hardware to write to a file or to send network packets to an internet address.
An intelligent agent is one that selects the ‘best’ action from the range available. In AI speak, the experts tell us that “for each possible sequence of input events, the intelligent agent selects an action that is expected to maximise its performance measure, given the evidence that has been provided by the input sequence and whatever built-in knowledge that the agent has”.
In simpler language, the intelligent agent assesses the available options and picks the best one, given its experience and knowledge of its surrounding environment.
Taking one step further, an intelligent agent is intelligent because it also has the ability to learn – to improve its performance over time.
One practical application of AI in asset management is in robotic process automation (RPA). This involves the use of robotics to process transactions, manipulate data and communicate with other digital systems. In practice, RPA uses software tools for repetitive processing and for completing standardised, low-complexity tasks without human involvement.
Natural language processing is the application of algorithmic models to analyse and understand human dialogue (written or spoken). Using this technique, the model will identify key terms from unstructured text. In the asset management industry, NLP is being applied in compliance software, for example, where it is used to identify keywords from unstructured text input and to populate this into structured data fields to enable audit, reconciliation and reporting functions. In another situation, NLP may be used to populate fields in a smart contract to support trading or reconciliation conducted on a blockchain.
Robo-adviser models make use of a range of AI techniques to offer investment advice and portfolio monitoring. Typically, the investor completes an online survey, which is used to evaluate their risk profile and return expectations. An algorithmic model will then select an investment strategy that aligns with these criteria. Investors can monitor the performance of their investments via an online dashboard and apply portfolio balancing in line with changes in investment conditions or their wealth management targets.
To complement such a strategy, AI techniques may be used to provide automated customer credit evaluation and to set credit limits. These techniques have been applied in a similar way by insurance companies for risk assessment and underwriting functions, providing guidance on the level of insurance cover that a customer should receive and the premium they should pay.
In customer service functions, asset managers are experimenting with using NLP and other AI techniques to process client queries. An intelligent agent may be used to interpret a standard client request such as updating their contact details or viewing their recent transaction history. If a request is more complex, the intelligent agent may refer this to the appropriate member of the customer service team for further action.
AI is also playing an important role in IT security and fraud detection. Usually this will involve running statistical analytics on input data to find patterns that lead to fraudulent activity (‘clustering’) and to identify users that present the highest risk.
Drawing on an intelligent agent’s ability to learn, AI plays an important role in identifying new patterns of fraudulent activity as criminals become more creative, while also minimising disruption created by false alarms.
For the next part of the report, click here.
©2019 funds europe