An in-person roundtable discussion on ‘Industry–Academic Partnership to Drive FinTech Innovation’ was organised by the Kotak IISc AI-ML Centre (KIAC), Indian Institute of Science (IISc) on 16 March 2023.
The aim of the roundtable discussion was to enhance cooperation between IISc and the FinTech industry, with particular emphasis on collaborative research in the areas of artificial intelligence (AI), machine learning (ML), and quantitative finance, in order to promote innovation within the financial technology (FinTech) sector. This is in accordance with one of the objectives of KIAC – the development of rigorous AI/ML-driven solutions to Fintech problems, in collaboration with industry and research laboratories. These solutions could then be hosted as open source or lead to deep tech start-up launches in the AI–Fintech space.
Delegates from Ernst & Young Global Limited, Talent Sprint, Fidelity Investments, NatWest Group India, Wells Fargo, State Street Global Advisors India Private Limited, and Indian Institute of Science were present for the discussion.
Faculty members from IISc indicated the possible routes for collaboration through scholarships and infrastructure support. They highlighted the exponential growth of data, a major contributor to which is the FinTech sector; the requirement of relevant and reliable data for models to assist in enhanced decision-making, system modernisation, and business transformation; the need for openness and trustworthiness; and the need for validation and reproducibility of the academically-generated synthetic data.
The possibility of recruiting interns from the Institute’s new undergraduate 4-year programme, BTech (Mathematics and Computing), was emphasised. The programme trains students in mathematical modelling as well as abstract thinking. The students have the flexibility of pursuing study tracks, based on their interest, such as mathematical finance, and AI & ML. In addition to the core and soft-core course requirements, the students must carry out projects in their seventh and eighth semesters; they are encouraged to carry out these projects in the industry. The students can also do short-term internships in the industry.
The industry partners were requested for responses to the following two main questions:
- What are the primary areas or issues that your organisation is currently prioritising for research?
- In what ways can your organisation facilitate collaborative research efforts with IISc?
The other questions that were posed included:
- Are there any India-focussed opportunities that we can work on?
- What do you think are the hindrances to collaboration with IISc?
- When we talk of FinTech, we talk of the ‘technology’ part but not much about the ‘finance’ part. What are the fundamental economic and financial problems?
- What are the immediate and long-term problems?
The key concerns were the following:
- How do we address the problem of data adequacy?
- How do we find clean data at the right time, given that coding is done differently by different people?
- There is a lack of historical data, to evaluate hypotheses in a robust manner. How do we do back data experiments?
- Can we meaningfully calibrate data? How do we handle unstructured data? How can we carry out synthetic data generation?
- How do we interpret macro data and use it in nano applications?
- How do we simulate something realistically?
- What are the modelling platforms that can be used? There is data in hundreds of pages of documents, which needs to be extracted. There is need for a system that can highlight model weaknesses.
- Artificial intelligence models are unable to identify structural shifts in economic regimes. Hence, human intervention will be required along with models.
- How do we manage open source projects using natural language processing (NLP) to capture the sentiments of different classes? Can we give and receive instructions in vernacular language voices?
- All AI NLP methods seem to be subject to alpha decay; the efficacy is from a few seconds to 2–3 weeks. This has to be scaled up.
- How do we map clients and their behaviour in order to better position sales and online social media efforts? Can we identify potential financial behaviour of customers, given the variety of data streaming in from different areas? Can we identify the same customer the next time?
- How do we attribute client results to the skills of our team rather than to the macro/market conditions (causal machine learning)?
- How do we tackle the churn problem? When high-profile advisors leave the company, they take with them a large client database…
- How do we obtain more insights into financial crime and fraudulent transactions?
- How do we assess credit worthiness?
- How do we define a fair organisation?
- How do we handle the environmental, social, and governance (ESG) factors with respect to climate change?
- How do we identify green washing?
- How do we ensure that our company is moving towards Net Zero from the customer aspect.
- Can there be a return on investment (ROI), where research can translate to on-the-ground delivery? What will be the application and implementability of research?
The results of the discussion led to some key action points:
- The next meeting should be more targeted/focussed; specific problems need to be selected for the next brainstorming session; suggestions from industry partners will be solicited through a form sent by IISc. A full-day workshop can be organised.
- An expression of interest needs to come from the company’s side, and then, the appropriate IISc faculty can be approached and invited to work on specific problems/challenges. There can be one-to-one connections and groups formed between IISc and the company.
- Certain groups can be formed between individual companies and IISc, and others can be a consortium that will work on common problems. These groups can meet on a monthly basis.
- Pilot collaborative projects can be taken up with very specific objectives, the scope of which can be expanded later. The problems can initially be taken up in small chunks of 3–6 month duration.
- Digital goods (intangible goods that exist in a digital form) is the next generation driver of AI/ML. Data has to be considered as an asset. However, there are no open source data sets in FinTech. There is need to have a repository of data from the industry. The Kotak IISc AI-ML Centre can help create such data sets, which can be used by researchers. Industry partners can suggest possible data sets. A regulatory framework needs to be formulated for digital assets.
- This collaboration between the FinTech industry and IISc can be a platform to voice opinions, share best practices, and highlight emerging technologies.