Banks’ AI: Is it more artificial or intelligent?
For the Week 5 post, I was hoping to get personal and blog about my own experience with AI in Finance. It is no secret that the last 24 months within the financial sector have seen an unprecedented surge of interest in Innovation and technology. Every global bank has implemented an innovation programme or opened their door to start up and external vendors in an unprecedented cultural move. Just this year, JP Morgan issued a 280-page report, titled “Big Data and AI Strategies: Machine Learning and Alternative Data Approach to Investing”. The report was branded as educational material to train asset managers on machine learning and data science. Goldman Sachs is also taking an interest: they recently issued an AI based investment trust to help investors make better decisions. These are a few examples. The interest has grown but it is still unclear if banks which mostly grew through years of mergers and acquisition, and consequently, developed patchwork architecture and infrastructure, can really innovate. I will say it depends on what part of the business and how innovation is tackled.
When it comes to AI a lot of the problems we face are common to any innovation venture but there are also some specific. Here is a few a list of some of the most familiar challenges:
1. Data is bad and compute capabilities outdated.
Due to the growth through acquisition data is siloed and definitions are not available or inconsistent. Also, most of the time data is incomplete. Data is also not well understood and therefore relations between variables are hard to explain. In what concerns compute most bank have not upgrade their capabilities to be able to accelerate AI.
2. Skill shortage
Shortage of data scientist that can understand, analyse and explain data in banks hinders progress. Also, AI engineers often opt to work for start-ups or big tech brands instead of banking so many financial institutions struggle to attract the talent.
3. Grand expectations
Whilst AI has existed for the best part of 60 years it has never received the hype it has gotten in more recent years. “According to Google Trends, only 5 percent of the U.S. population searched for information about artificial intelligence in 2012. According to Google Trends, only 5 percent of the U.S. population searched for information about artificial intelligence in 2012.” Misunderstood potential can often mean, ambitions are very short lived.
4. Budgeting cycles and heavy governance
Most banks operate a yearly budgeting cycle that take ups approximately 1 quarter. For that reason, any multi-year projects always suffer from delays and cost inefficiencies as they must shed resources at year end and then ramp up at start of year. Also, a lot of valuable high-cost resource time is spent justifying spend to committees for approval.
5. Banks rely on their full-time staff to come up with use cases
Most banks rely on the same staff that have a 9 to 5 job to come up with ideas and pursue use cases on their own. Often that is also accompanied by lengthy internal vendor management procedures and a multitude of reporting to various senior teams. As a result, of poor incentives and time allocation, it is not possible for meaningful progress to really happen.
6. Only Cost Saving innovation counts as innovation
Often the only criteria management relies on to approve budget, is how much saving the use case will contribute i.e. how many Full-Time Employee can be fired. This inhibits a lot of experimental innovation typically found in AI and restricts innovation to automation and robotization (Robot Process Automation). It encourages employees to also not come forward with ideas so they do not compromise on job protection.
To conclude:
For an organisation to innovate, it needs to minimise barriers to entry and implement incentive programmes.
These incentive programmes do not have to be associated with financial remuneration as much as, they should allocate time for resources to spend effort on innovation and acknowledge contribution made even if use cases fail or do not deliver the expected outcome. This is very important as divergent thinking, means thinking up many solutions to the same problem as opposed to pondering over the one optimal solution. Budget approval should be done on an incremental basis (to minimise cost) and should be based entirely on the potential of the use case as opposed to headcount saving. Specific to AI, banks should also focus on getting their data and compute capabilities right before rushing into applications. Finally, internal governance and red tape should be regularly reviewed and reassessed against value it provide and risk it mitigates, to ensure that “more” doesn’t become “less”.
Bibliography
Avoyan, H. (2017, May 10). Why the AI hype cycle won’t end anytime soon. Retrieved from Venture Beat: https://venturebeat.com/2017/05/10/why-the-ai-hype-cycle-wont-end-anytime-soon/
Critchfield, S. (2017, March 09). How to Push Your Team to Take Risks and Experiment. Retrieved from Harvard Business Review: https://hbr.org/2017/03/how-to-push-your-team-to-take-risks-and-experiment
JP Morgan: Alternative Data Is Altering Investment Landscape . (2017, June 14). Retrieved from Integrity Research: http://www.integrity-research.com/jp-morgan-alternative-data-altering-investment-landscape/
The Use of AI in Banking is Set to Explode. (2017, Jan 17). Retrieved from The Financial Brand: https://thefinancialbrand.com/63322/artificial-intelligence-ai-banking-big-data-analytics/
Users who have LIKED this post:
2 comments on “Banks’ AI: Is it more artificial or intelligent?”
Comments are closed.
Hi Lana and thanks for an interesting read!
You definitely listed some valid challenges banks will face when implementing AI into their processes. One thing that I think you could’ve talked a bit is the role of regulation; just like healthcare, I believe finance has been slow to reinvent itself due to the slow-moving legislative frameworks.
On the other hand, what I find particularly interesting is which functions and areas of the big banks will be replaced or disrupted by AI or tech generally. Certainly many parts on the markets-side (typically pricing of financial instruments) of banking have been data-driven for a long time already and discussed in various articles, but the advisory-side (M&A, etc.) has been more recently in the news for initiatives to automate some of their core functions. The big investment banks’ advisory businesses have been known to be very labor intensive, but also too challenging for a typical ML-algorithm to automate, which is why I believe it could “leapfrog” directly to the AI-era of professional services (at least a few investment banks are already automating some of their basic services for their people to focus more on customer relationships).
Great post! In the context of technology and innovation in finance I couldn’t help but comment about my experience in Egypt and think about cryptocurrencies. All around the world globalization, capitalism, governments and even democracy are in retreat while populism, nativism, protectionism are on the rise. People have less and less trust towards institutions, governments and have recently started awakening to the power they have been giving away to large corporations.
Another important fact to mention is that a large part of the global population is still excluded from services that are considered “basic” today. In our open and digital economy, 39% of the world population don’t have a bank account, thus don’t have access to financial services. Especially in my home country, Egypt, where only 7% of adults have bank accounts. Although those populations usually have access to Internet (mainly through mobile phone), the globalization and internationalization of our economies still couldn’t (or didn’t) reach them. Some will argue it isn’t profitable enough for large banks to provide financial services to a population whose total transactions can’t be counted in billions of dollars. The end fact being that a large proportion of today’s world is not included in the economic world because of a lack of integration from the service providers.
In this context of rising awareness and need for more global solutions, the blockchain technology benefited from a great momentum. It opened up doors for new opportunities, using cutting-edge technology to serve society. The technology initially underlying the Bitcoin protocol is said to have the power to have a great impact on our society by transforming our industries. From financial services to healthcare services, including governance, a wide range of fields could be disrupted using the blockchain technology. And by disrupting those multiple fields, it could also allow for a better inclusion of the significant population that globalization left behind. If we can now transact over the internet without the need for a trusted intermediary, the possibilities for innovation are limitless. Although it has its limitations, society as a whole can benefit in many ways from the global adoption and establishment of such a technology.