Does your Bank’s AI Partner have The Right Data Strategy?
It’s easy to fall into the trap of expecting the next wave of tech innovation to resemble the most recent one. When the iOS App Store opened its doors to developers more than a decade ago, many entrepreneurs made millions building products on laptops with just a little code.
It would be a mistake to think that the AI revolution will be this easy. Training a great algorithm — the AI equivalent of building a great app — isn’t enough to create a successful AI initiative at your bank. An algorithm is only as good as the data it’s trained on. And data management and governance in financial services is extremely complex, especially with data privacy laws like GDPR and CCPA now on the books.
Adding to the complexity, it’s become increasingly clear that the easiest way for banks to access innovation in AI is to collaborate with fintechs. While fintechs familiar with the industry will have standard protections in place, from the bank’s perspective, sharing data with any outside partner increases complexity.
Why AI Needs Data Strategy
In graduate school, AI researchers don’t have to worry too much about sourcing, securing, or managing the data they use. They’re often working with large, publicly available datasets that have been anonymized to protect privacy. In academia, it’s possible to say “give me the data, and I’ll do my work.”
The AI landscape in financial services is different, particularly when it comes to bank-fintech partnerships. Banks aren’t going to hand over proprietary datasets that are core to their value propositions without strong assurances that data will remain secure and that customers’ privacy will be protected. Even with those assurances, banks may feel wary about co-locating datasets on a third-party vendor’s servers. Moving data outside a bank’s firewall always increases risk, no matter what protections are in place.
Banks and fintechs that don’t address these issues will struggle to scale AI in a meaningful way or derive real value from their algorithms. Any bank that wants to be part of the AI revolution will need to develop a strong data strategy — a plan for how they will source, manage, govern, and secure data throughout their organization.
Any partner a bank works with will need to fit into that strategy and also provide their own plan for how they’ll handle the data the bank entrusts them with. Without a data strategy in place, a fintech will struggle to find partners and therefore to scale.
How Tokenization Protects Privacy
A fintech’s data strategy should include robust data governance and management policies that reassure banks their data is safe after it leaves the protection of their firewalls. However, with data tokenization it’s no longer necessary for sensitive data to pass beyond those firewalls at all.
Tokenization replaces sensitive data with non-sensitive representations (tokens) that can then be used for analysis without violating privacy or revealing proprietary information. By analyzing data tokens instead of the data itself, an AI startup can generate insights from sensitive data without increasing risk of a breach.
It’s worth mentioning that data tokenization can enable secure data-sharing with all sorts of other partners, too. Combined with powerful decentralized technologies like blockchain, tokenization will reinvent the data ecosystem as we know it by enabling data-sharing alliances that span multiple industry verticals.
However, while tokenization is industry-agnostic, data strategies shouldn’t be. Data management best practices vary widely from industry to industry and from niche to niche. In financial services, strict regulations make data governance incredibly complex, demanding specialized solutions. Even IBM Watson — probably the most powerful, general-purpose AI engine out there — has built out AI products and services specific to the financial sector. Whether using tokenization or not, banks’ and fintechs’ data strategies should include detailed plans for addressing the unique needs and pain points in the industry.
The Future of Data Strategy in AI
Tokenization probably isn’t the be-all, end-all when it comes to ensuring privacy and security within AI applications. For example, if someone figured out how to train machine learning algorithms on encrypted datasets at scale, it would create a new multi-million-dollar industry almost overnight. It would also completely change how the AI applications used by banks and fintechs ensure the security and privacy of partner data used for analysis.
But that’s in the future. Today, tokenization is already preserving privacy and security in AI applications. And for many banks, the right data strategy will be a vital component for leveraging AI effectively.