To ensure competitive advantage and future business success, financial services firms are scaling artificial intelligence (AI) and machine learning (ML). Those who succeed can vastly increase their return on investment. Those who don’t risk losing their market share as AI/ML scale, insight and efficiency propels their competition ahead.
- Financial firms are scaling AI and ML to manage risk and uncover new insights across business units – Refinitiv’s 2020 AI/ML report shows 66% of respondents using ML for risk, 31% for investment research, 30% for customer service, and 27% for trade execution.
- AI/ML at scale requires precise execution capabilities which favors a three-pronged approach to – 1. breadth of talent, 2. depth of technology tools, and 3. diagonal expansion of apps and use cases.
- Change is exponential in the AI/ML space, firms that fail to scale successfully risk being unable to catch up once the financial services industry has surpassed an adoption tipping point.
Refinitiv’s second Artificial Intelligence (AI) / Machine Learning (ML) Report, ‘The rise of the data scientist: Machine learning models for the future’, shows financial services firms are no longer treating AI/ML initiatives as experimental proofs of concept (PoCs) and are instead using these technologies in fully-fledged production environments optimized to scale across business units, geographies and use-cases.
The driver behind this hunger for AI/ML is the aim to uplift revenue with new alpha generating opportunities across the business, cutting down costs by automating manual processes, identifying risk quickly and accurately, and gaining a competitive advantage over fellow player in the market who may not be moving as fast.
According to research by Accenture, firms that scale successfully see three-times the return on AI investments compared to those who are stuck in the PoC stage. Some 84% of C-suite executives believe they must leverage AI to achieve their growth objectives, and three out of four believe that if they don’t scale AI in the next five years, they risk going out of business entirely.
The 2020 Refinitiv AI/ML report – one of the leading and largest financial AI/ML research studies in the world covering over 400 data scientists, quants and tech leaders in the Americas, Asia-Pacific and EMEA – is built on an initial report carried out by Refinitiv in 2018.
Three steps to scale AI/ML
Contrary to popular belief, scaling AI/ML is not just about building larger models that can consume higher volumes of data. It’s a complex execution challenge that requires a three-pronged approach including broader data science skill sets, deeper tech capabilities and diagonal expansion across the business.
Step One: Data scientists add breadth
Breadth can be achieved by deploying more data scientists in executive roles and empowering them to lead the full stack of a firms data strategy, including data acquisition, AI/ML tool deployment and emerging tech talent sourcing.
This year’s Refinitiv report recognizes the growth of the data science community – both in size and influence – in financial services. The number of teams within firms has seen an exponential increase of over 260% since 2018, and data scientists are now more likely to be found working across different business units, rather than as part of a support function in technology.
Citigroup plans to hire 2,500 programmers and data scientists for its trading and investment banking units. Similarly, HSBC has set up a quantitative finance associate programme in the UK.
Refinitiv helps financial services firms underpin this breadth of scaling AI/ML with innovative solutions like its data exploration tool – built by data scientists in Refinitiv Labs for data scientists across finance. A single interface hosts a collection of Refinitiv datasets, services, documents and pre-built Jupyter notebooks, ready for fast exploration and experimentation across a number of use-cases, including alpha generation, trade execution and risk management.
Step Two: Technologies add depth
Depth in scaling is reflected in how financial firms’ of 2020 specialize their use of AI and ML tools, invest in emerging technologies and double down on turning their data into robust sources of business intelligence by leveraging multiple cloud providers to ensure resilience.
An explosion in the use of deep learning, 75% of firms using the technology to drive high value outcomes with their AI/ML strategy, has created a surge for the latest frameworks and tools in the market.
Taking inspiration from tech giants like Google and Facebook, firms are adopting open source tools to scale their internal legacy systems before the paradigm shift in technology requires a complete overhaul vs a relatively cost efficient business operations lift and shift.
Depth of AI/ML scale is supported by the Refinitiv Data Platform, a new cloud-based data experience with unmatched depth of data coverage. Its data access options allow companies to meet their organization’s permissioning needs, while providing a high quality, consistent, clean and normalized data feed for AI/ML applications.
Step Three: Scaling diagonally
Diagonal scaling increases the number of apps and use cases of AI/ML across the business. The strategic thinking behind this pillar is to find creative ways to deploy the same AI/ML tool or model across multiple business units, use-cases or markets in spaces such as risk, trading and investment. Both for today and future expected objectives and constraints.
This dimension of scaling also applies to data, where firms are expanding their data repertoire to include unstructured data and the required advanced insight extraction techniques such as Natural Language Processing (NLP) to amplify apps and gain competitive advantage. Alternative data is one such high-growth data category which developing rapidly to meet market demands as an alpha generating secret weapon.
One example of how Refinitiv ensures its data powers scaled AI/ML models that achieve accuracy and precision is PermID, or Permanent Identifier. PermIDs are open, permanent, and universal identifiers for Refinitiv data. They are machine-readable and help data practitioners handle complex data management challenges by eliminating mapping inconsistencies, reducing operational risk, and streamline end-to-end workflow processes.
Barriers to scaling
The problems of data quality and availability identified as barriers to AI and ML adoption in our first report in 2018 persist, and have become a greater challenge in 2020. At the same time, issues around talent, funding and technology are fading as the AI/ML ecosystem is growing to critical mass where the sheer number of practitioners innovate and battle test new tools, prove their business value and lock in funding.
While we have not crossed the chasm of AI/ML adoption yet, as seen in the persistent shortfall in experience of developing AI/ML apps that continues to hamper many firms, there are now many firms that have leaped from experimentation and PoCs to build production grade solutions that can be scaled, with solid budgets, talent and tools in place to ensure the initiatives long-term success.
Benchmark your AI/ML maturity and readiness to scale
Download the full Refinitiv 2020 AI/ML report, which includes global and regional insights into:
- How firms are scaling AI/ML across multiple business units.
- The role and increasing influence of data scientists.
- The most popular frameworks and platforms.
- Key challenges to overcome.
- Acceleration and investment post-COVID-19.
- 2021 predictions for AI/ML in finance – including the rise of MLOps and explainability.
