L’ESA sécurise ses développements informatiques terrestres avec Telindus

Author: Sebastien Lambotte
18/01/2018
Cybersecurity

AI for Finance? Sure, but the right one!

 

Among the trending FinTech topics, Artificial Intelligence (AI) beholds a prominent position, having re-emerged from a few winters. This new comeback is based on a set of factors: 1.) data, that has been subject to explosive growth; 2.) technology leaps in fields of pattern recognition and computing power; 3.) importantly, user benefits that have emerged.  

 

The Financial Sector is mainly driven towards AI by the anticipated boost in competitiveness. To achieve this, the various AI technologies and their applications must first be evaluated in view to define their strategic congruence. Indeed, one must realize that existing AI tools are not holistic but represent single or combined capacities like for example Problem Solving, Learning, Information Processing, Manipulation, Social/Emotional Interaction, Projections, Decision, Creativity, etc. As there are many AI use-cases in Finance that differ in terms of their objectives as well as their targeted users, it is important to select the right mix of AI tools.

 

Robo Investment Advisors, for example, monitor the investment universe in view to find assets fitting a given client’s capital planning. They will do this by combining Information Processing, Problem Solving, Projections and Decisions.

Robo Portfolio Managers, on the other hand, are pre-selecting investments for Investments Professionals that conform to a given investment strategy by pouring over and analyzing formalized information (e.g. annual reports) plus unstructured data (e.g. web search engines) at record speed and then pick using Decision AI.

Robo Underwriters that automate the insurance and lending process, will rely heavily on Decision capacity in view to make better & faster decisions. By adding Manipulation, the client can then directly receive a tailored offer.

New-generation ChatBots, rely heavily on Social/Emotional Interaction, that enables them to read the emotional content of client queries and to communicate the AI’s findings with the right level of empathy.

FraudBots meanwhile, greatly depend on Information Processing and Learning in view to identify unusual patterns and their evolution.

Interestingly, the Forensic Analysis by Financial Regulators of past actions undertaken by supervised entities, is now being explored with a blockchainized version of AI that builds on Learning, Information Processing.

 

But the specific challenges linked to AI in Finance should also be in integrated into AI implementation.

Essentially, the quality of the data and the initial matrix, demand ongoing scrutiny as they form the core of any AI. Here tools to trace and measure absolute and relative outcomes in financial terms must be installed.

AI’s high-speed and independent task-execution entail the risk to induce significant damage before any glitch is noticed. Therefore, pre-set circuit-breakers and/or the kick-in of human supervision are required in order to contain contamination and ensure systemic resilience. An AI on the AI in a sense.

As regulation is outdated or simply inexistent when it comes to AI, it is quite difficult to ensure compliance. The similar concern applies to legal aspects, especially on the liability side, for which there is a high level of uncertainty regarding who would be responsible if anything goes wrong: The regulated institution, the AI manufacturer or the AI maintainer? Here a sound AI risk governance is required.

User-experience, is also a key factor as the quality of interaction by and with users determines the effectiveness of AI tools. Here, an all-important aspect also resides in having the right HR capabilities to implement, monitor and manage AI by staff mastering both AI-Technology as well as Finance.

 

Once the right AI-mix has been defined it is important to prioritize based on segments and scope. Should the focus lie on productivity, like for example with a retail bank, where AI can lower costs and increase execution speed? Or should the emphasis be on margin-generation through the design of a unique client experience, as with Ultra-High-Net-Worth individuals for whom AI becomes an omni-accessible Personal Financial Assistant? Regarding the scope, should AI be first applied to single task that are low-hanging fruits or immediately to broader value-chain segments?

 

Finally, an essential point consists in humanizing AI as only ergonomic and collaborative forms of AI will be embraced by clients and professionals alike. Clearly, the successful tandem between humans and AI will generate lasting added value…

 

by Frank Roessig- Article for SocGen Magazine

Head Digital Finance Solutions at Telindus (Proximus Group)

Une question ?

 Contactez-nous ! Nos équipes sont à votre entière disposition. Quelle que soit votre problématique, nous trouverons une solution ensemble.

Enter the characters shown in the image.
Cette question sert à vérifier si vous êtes un visiteur humain ou non afin d'éviter les soumissions de pourriel (spam) automatisées.
By submitting this form, I accept the Telindus personal data protection
* mandatory fields