Seldon has been working at the intersection between machine learning and fintech, and was part of the Barclays Accelerator London ’16 cohort. This perspective has given me a chance to observe some nascent trends for AI and machine learning in fintech that will define this year:
- AI in Production – AI is only used by banks in production in a few key use cases such as high-frequency trading, fraud detection and credit scoring. In 2016 many machine learning R&D projects started across other business functions. In 2017 banks will move from testing machine learning models to putting models into production to make a real impact on business KPIs.
- Open-Source AI Platforms – Leading on from the last point, banks will have to consider if the best strategy for operationalizing models is to use a major cloud vendor, proprietary tech, open-source tech or in-house build. I think the winning combination is an open-source core machine learning platform supported by in-house R&D higher up the stack, and cloud provider focused mostly on the lower level compute tasks.
- Open Data – The new EU payments directive (PSD2) compels banks to open up customer accounts data to third parties authorised by the customer. Access to richer datasets, which were previously a barrier to entry, creates opportunities for fintech to build better models, leading to more intelligent apps and services.
- Data Security 2.0 – Banks will upgrade data access and extraction procedures. Techniques such as differential privacy and homomorphic encryption will enable secure end-to-end data access for internal and external projects and faster innovation cycles.
- AI Regulation – Banks and fintech companies will work alongside regulators to make clearer regulations that remove grey areas and speed up adoption and innovation. Some advanced techniques such as bias prevention, auditable and explainable of machine learning models will be essential for AI projects that move to production.
- Decision-Making – Automation will continue to progress from simple rules-based systems to complex augmented and autonomous decision-making systems. The Connecticut-based hedge fund Bridgewater is leading the charge as it seeks to replace more managerial functions with AI. Traditional banks will start experimenting with the same concepts, initially with smaller projects in isolated business units.
Closer Collaboration – Banks will gain a competitive advantage by innovating faster in partnership with fintech start-ups. It’s a learning experience for all, and with each iteration processes are streamlined to make future projects more effective. Fintech accelerators will continue to grow and play a key role in connecting banks with fintech start-ups. Techstars Barclays Accelerator the gold standard that will continue to go from strength to strength in its new Shoreditch co-working space that connects alumni and members of the fintech community.
Closer Collaboration – Banks will gain a competitive advantage by innovating faster in partnership with fintech start-ups. It’s a learning experience for all, and with each iteration processes are streamlined to make future projects more effective. Fintech accelerators will continue to grow and play a key role in connecting banks with fintech start-ups. Techstars I’m excited that Seldon will play a part in some of these trends. To find out more about our machine learning deployment platform, check out our technical docs and open-source project.
If you have any feedback or questions, we’d love to hear from you — contact [email protected].
Here’s to a great 2017 for AI in fintech!