We are seeing increasing examples of machine learning in many areas of technology and financial institutions should grasp the opportunity to use it for repetitive analysis. There are also alerts that are generated which carry a relatively low money laundering risk, e. g. an alert based on a counterparty moving funds between their own accounts (possibly held in different jurisdictions or in different banks). False positives are, however, still very common for most algorithms. If the system could perform a simplified up‐front analysis to exclude these types of alerts, or move them to a specially quarantined area, then allied with self‐learning, the alert landscape becomes cleaner and only the truly suspicious transactions require costly human intervention.
31.5 Challenges to Artificial Intelligence Deployment
From the industrial revolution to the emergence of the Internet age there have always been challengers to the deployment promoting fear of a change in traditional roles and jobs being lost. AI adoption is very much in the eye of the storm with its perceived ability to provide reliable results from large data sets and to carry out cognitive and creative tasks more competently than humans.
31.5.1 Consumer Data Protection and Privacy Challenges
The use of big data raises a number of privacy and data protection concerns, including transparency (what data are being used and where did it originate) and customer consent (was permission provided for its use and possibilities of access/redress, can the consumer see their own data and request that errors be corrected). While these issues may be addressed for the data in credit bureaus – which originate with banks and other formal lenders and service providers – the same protections do not typically extend to big data that are amassed from a combination of private commercial transactions, government sources and publicly available information such as social media posts. Developing a practical approach to consumer protection for big data, which balances privacy and consumer protection with commercial applications that can facilitate commerce and even access to credit, is a challenge that remains to be met.
31.5.2 Automation of White Collar Jobs
AI is having an impact not just on routine and repetitive tasks but also on cognitive, and even creative tasks, as well – see for example [22] where authors used Generative Adversarial Networks to generate faces and bedrooms. A tipping point seems to have been reached, at which AI‐based automation threatens to supplant the brain‐power of large swathes of middle‐income bank employees. The example raised earlier of RBS of releasing 550 staff, to be replaced with ‘robot‐advisers’, is evidence that AI will have a tremendous impact in workforce by rendering many jobs obsolete [7].
Not only is AI software much cheaper than mechanical automation to install and operate, there is a far greater incentive to adopt it—given the significantly higher cost of knowledge workers compared with their blue‐collar brothers and sisters in the workshop, on the production line, at the check‐out and in the field. The impact is real as far few new white‐collar jobs are expected to be created to replace those at risk of being lost—despite the hopes many place in technology, innovation and better education.
What constitutes work today—the notion of a full‐time job—will have to change dramatically. The things that make people human—the ability to imagine, feel, create, adapt, improvise, have intuition or act spontaneously—may remain comparative advantages with respect to machines, but even those are not guaranteed as the processes where these competencies emerge can themselves be described by algorithms [1]. The most recent advanced in AI proved powerful enough to beat the best human players in the challenging game of GO [23] – something regarded as impossible a few years ago as it requires the machine to learn high level strategy – though to be an human exclusive capability.
There is another flipside of AI. The fact that Deep Learning algorithms are probabilistic cognitive machines controlled by million of parameters, makes them almost impossible to be understood. This is particularly restrictive for banks as, for instance the scoring algorithms, have to be white boxes, due to regulation requirements. Furthermore, other risks may emerge as a consequence of relying on “black box” algorithms: how can we be sure they less biased than humans? How will they cope with novelty or non‐stationary data? How can we be sure they are reliable?
We should be prepared for smarter than human systems acting in reality that may encounter situations beyond both the experience and the imagination of the programmers – see [1] for further exploration of these possibilities.
31.6 Conclusions and Recommendations
Despite the technological threats, Deep Learning powered AI is transforming businesses at an accelerating speed. According to IDC, by 2020, the market for machine learning applications will reach $40 billion. In 2015, only 1% of software include AI, in 2020 this number will be above 50%. In the future, AI will fundamentally change and automate innumerous functions within companies: from pricing, budget allocation, fraud detection and security to marketing optimization if organizations implement it in the right way.
Machine Learning will make everything in the organization programmatic, from advertising to customer experience. This allows companies to create products that interact naturally with humans and will force reorganize several departments in financial institutions.
Deep Learning is well suited for activities that are heavily data intensive, like advertising and click‐through information. Most of the data will be collected by the mobile phones and a myriad of devices delivering real time geo‐referenced information. Multimodal learning will allow the integration of text, images, video and sound within a unified representation.
Banks need to exploit the opportunity of digitalization and advanced algorithms capable to radically change their business. Banks have deployed online servicing, capacity‐management software, interactive voice response systems but they’re not using them widely and still dragged by inertia. They also need to avoid the trap of deploying AI purely for cost saving initiatives and build forums that could potentially stifle innovation. Banks need to take positive lessons from the large IT firms open up their APIs and adopt a more customer centric approach allowing the real innovators to innovate with more freedom to enable the rapid turnaround of exciting applications.
Too often, banks manage the progress of their digital transformations by tracking activity metrics, such as the number of app downloads and log‐in rates. Such metrics are, however, inadequate proxies for business value. Banks must set clear aspirations for value outcomes, looking at productivity, servicing‐unit costs, and lead‐conversion rates, and link these explicitly to digital investments. Additionally, deeper awareness of the technical capabilities available and how they can affect processes will be a prerequisite to effectively manage in this new world.
We are on the verge of the biggest revolution of all times. Banks can take huge opportunities in assuming digitization will change the traditional retail‐banking business model, in some cases radically. The good news is that there is plenty of upside awaiting those European banks willing to embrace it. The bad news is that change is coming whether or not banks are ready.
This disruptive technology do not come without risks. The relentless data stream that no one can cope but the inscrutable algorithms that process it, can have consequences that no individual can plan, control or comprehend. Maybe that no one really needs to understand as long as it increase the profitability and loyalty of customers. Users, however, will be more preoccupied by the lack of privacy, The risk of data‐processing system becomes all‐knowing and all‐powerful, so connecting to the system becomes the source of all meaning. This concentration of cognitive computing in a handful of players will create a threat to banks.
References
1.
Y. N. Harari, Homos Deus, Harvill Secker, 2016.
2.
> B. Agarwal, S. Ambrose, S. Chomsisengphet und C. Liu, The role of soft information in a dynamic contract setting: Evidence from the home equity credit market, Bd. 43, J. Money Credit Banking, 2011, p. 633.Crossref
3.
J. Schmidhuber, “Deep learning in neural networks: An overview,” in Neural Networks, Bd. 61, 2015, pp. 85–117.Crossref
4.
I. Goodfellow, Y. Bengio und A. Courville, “Deep Learning,” MIT Press, 2016. [Online]. Available: http://www.deeplearningbook.org.
5.
A. Vieira, Deep Neural Networks: review and business applications, Apress (in preperation).
6.
A. Krizhevsky, I. Sutskever und G. Hinton, “Imagenet classification with deep convolutional neural networks,” in Advances in Neural Information Processing Systems, Bd. 25, Curran Associates, 2012, pp. 1106–1114.
7.
The Guardian, “RBS to cut 550 jobs as part of plan to automate investment advice,” 03 2016. [Online]. Available: https://www.theguardian.com/business/2016/mar/13/rbs-royal-bank-scotland-cut-550-jobs-automating-investment-advice.
8.
M. Lin, N. Prabhala und S. Viswanathan, Judging borrowers by the company they keep: Friendship networks and information asymmetry in online peer-to-peer lending, Bd. 59, Manag. Sci., 2013, pp. 17–35.Crossref
9.
S. Agarwal, S. Chomsisengphet, C. Liu und N. Souleles, “Benefits of relationship banking: Evidence from consumer credit markets,” J. Social Sci. Res. Netw., Rochester, NY, USA, 05 2009. [Online]. Available: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1647019.
10.
Convolutional LSTM Network, “A Machine Learning Approach for Precipitation Nowcasting,” 2015. [Online]. Available: http://arxiv.org/pdf/1506.04214v2.pdf.
11.
Finextra, “Deutsche Bank crowdstorms AI ideas,” 05 2016. [Online]. Available: https://www.finextra.com/newsarticle/28855/deutsche-bank-crowdstorms-ai-ideas.
12.
Finextra, “Absa to trial AI-driven chatbots to answer customer queries,” 04 2016. [Online]. Available: https://www.finextra.com/newsarticle/28794/absa-to-trial-ai-driven-chatbots-to-answer-customer-queries.
13.
McKinsey Global Inst., Big Data, The Next Frontier for Innovation, Competition, Productivity, New York, NY, USA, 2011.
14.
J. Neves und A. Vieira, Improving bankruptcy prediction with hidden layer learning vevtor quantization, Bd. 15, European Accounting Review, 2006, pp. 253–271.Crossref
15.
A. Kracklauer, D. Mills und D. Seifert, Customer management as the origin of collaborative customer relationship management, C. C. R. M. T. C. t. t. N. Level, Hrsg., Berlin, Germany: Springer-Verlag, 2004.Crossref
16.
E. Ngai, L. Xiu und D. Chau, Application of data mining techniques in customer relationship management: A literature review and classification, Bd. 36, Exp. Syst. Appl., 2009, pp. 2592–2602.Crossref
17.
O. Vinyals und Q. Le, “A Neural Conversational Model,” 2015. [Online]. Available: https://arxiv.org/pdf/1506.05869.pdf.
18.
Kasisto.com, “ai-driven-virtual-assistant-from-kasisto-powers-indias-first-mobile-only-bank,” 04 2016. [Online]. Available: http://www.kasisto.com.
19.
Indian mobile-only bank handles customer service with chatbots, 2016. http://www.forwardlook.com/indian-mobile-only-bank-handles-customer-service-with-chatbots/
20.
techncrunch.com, penny-is-a-chat-based-personal-finance-coach, 2015.
21.
do-your-banking-with-a-chatbot, Massachusetts: MIT Technology Review, 2016.
22.
A. Radford, L. Metz und S. Chintala, “Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks,” corr/RadfordMC15, 2015. [Online]. Available: https://arxiv.org/abs/1511.06434.
23.
nature.com, 2016. [Online]. Available: http://www.nature.com/news/google-ai-algorithm-masters-ancient-game-of-go-1.19234.
24.
B. Ribeiro und N. Lopes, “Deep belief networks for financial prediction,” in International Conference on Neural Information Processing, 2011.Crossref
Part VIII
Fin- & Insuretech
© Springer-Verlag GmbH Germany 2018
Claudia Linnhoff-Popien, Ralf Schneider and Michael Zaddach (eds.)Digital Marketplaces Unleashedhttps://doi.org/10.1007/978-3-662-49275-8_32
32. Preface: Fin- & Insuretech
Hartmut Mai1
(1)Allianz Global Corporate & Specialty SE, Munich, Germany
Hartmut Mai
Email: [email protected]
32.1 The Landscape
32.1.1 There Is Nothing New About Technology
Technology and finance have gone hand in hand for decades, if not centuries. The insurance underwriter who has been abreast with the new trends was usually the underwriter who made the largest profit. That knowledge was based mainly on having the best contacts and the best technology.
When communications improved dramatically in the 19th century, insurers did not launch their own telegraph companies. They knew that this was something that other companies could do best. However, astute insurers realized that speedier communications and more efficient technologies would have a profound impact on their business.
All parts of the insurance value chain can become more efficient by understanding the potential of new technology, and exploiting it. You don’t need to know precisely how a car works, but it helps if you know how to drive one. The same applies to InsurTech today. Insurers and their staff do not necessarily have to understand the technical details of a new process, but they do need to know how to exploit it to its utmost potential.
32.1.2 The FinTech Investment Universe
According to US‐based Route 66 Ventures [1] (see Fig. 32.1), the FinTech sector – including InsurTech – is being serviced by three types of companies: the existing financial services groups, the FinTech service and product providers, and the Fintech disruptors. Those three types of companies are all working on back‐end, middleware and front‐end services.
Fig. 32.1US‐based Route 66 Ventures. (Accenture [1])
In turn, these operate in the business to business, the business to consumer and the so‐called B2B2C spaces. And bridging and linking these spaces and companies, within this we have a distinct dimension of its own: the highly important “protocol” dimension. Blockchain is the highest profile example of this area of FinTech, which increasingly extends both outside, and deeply within, every part of the sector.
Finally within the sector, we have the five FinTech core business sectors, namely Lending, Payments, Capital Markets, Asset/Wealth Management and Insurance, It is these in which the incumbents, the service providers and the disruptors are looking to operate.
32.1.3 FinTech: A Short but Exciting History
The mass infiltration of technology that we know of as “FinTech” – and its insurance subset, “InsurTech” is about five years old. It did not suddenly appear out of nowhere. Extraneous events – the financial crisis of 2008/09 and the subsequent collapse in interest rates – served as catalysts. The banking crisis showed that things could not go on as they had before, and the plummeting of interest rates forced insurers to face the fact that there was a “new normal”.
Global investment in FinTech grew from $930 m in 2008 to more than $12 bn in 2014 [1] and to $19.1 bn in 2015 [2].
Because banking suffered a bigger capital crisis than did insurance, banks had to respond faster. Accenture’s report on “The Future of FinTech and Banking” [3] identified three key responses necessary for a financial operation to survive and thrive, to turn threat into opportunity.
> They were: 1)Engage in a transparent way with technology solutions providers, permitting an interchange of intellectual property and data with chosen outside innovators;
2)Collaborate with partners beyond the financial services industry, partners who can provide a new perspective and who would not tend to travel the obvious path;
3)Invest, using venture investment to source new technology, to the benefit of the insurer.
This does not mean that each part of the value chain in insurance needs to employ all three strategies, but it does mean that insurers, if they are to fully exploit the potential of InsurTech, must devise strategies that will work in the short‑, medium‐ and long term.
32.1.4 FinTech Today – As at September 2016
FinTech is continuing to branch outward beyond the traditional payments and lending space, possibly because the “obvious” FinTech areas are now rather crowded with startups. InsurTech, Asset Management (Robo‐Advisory) and Capital Markets offerings have become increasingly popular in the past 18 months. “RegTech” companies have been offering companies automated systems to ensure that they are in compliance with regulations in various jurisdictions. InsurTech is just one of eight key verticals – among them wealth management, payments, blockchain, mobile banking or financial data – listed in the KPMG report on the Pulse of FinTech [4].
Digital Marketplaces Unleashed Page 46
