AI and tech investments need to show results beyond the CIO's office

sinenkiy/Angelov - stock.adobe.com

Artificial intelligence was the talk of the American Banker Digital Banking conference held this week in Boca Raton, Florida, (where McKinsey was the knowledge partner). The buzz is easy to understand, given its potential to boost productivity and lift so many elements of banking today to a new level, particularly in the context of several sectoral headwinds. For CIOs and C-suites thinking about digital and AI, the well-known challenges of deploying AI (e.g., risks like hallucinations and bias, and the difficulty of scaling pilots) are compounded by three major factors: the need to demonstrate a return on investment, or ROI, on past technology investments; the need to differentiate the bank from competitors; and the need to achieve success in their existing transformation efforts.

To date, our industry's record at these is uneven. Demonstrating the ROI on tech investments is not simple — indeed, for the industry, higher revenue remains very strongly correlated with more manual work. If tech were truly resulting in automation, we should be seeing significant returns to scale, but these appear absent from the data. Factor in that a lot of the spending has been (justifiably) on infrastructure modernization and risk management which does not necessarily generate revenue. Moreover, truly separating from the pack is difficult given how quickly innovation spreads through the industry. Additionally, our research finds that a mere 30% of transformation initiatives fully succeed.

Thus, the big challenge for banks in the years ahead is how to deliver material outcomes from their spending on tech, now including AI. That means not just delivering products on time and on budget but doing so in a way that generates measurable value — by generating revenue, removing cost or tangibly improving risk management, among other benefits. Apart from regulatory prudence, this may be a reason that while many banks have made initial gestures at testing AI, only a few have yet put real heft behind it.

A key insight that our research has found is that capturing value from technology and AI requires taking actions beyond just those domains. For example, in surveys we have conducted, 60% of executives cite skill gaps as an obstacle that they have had to address in their digital transformations, and 70% say they faced fundamental resistance to change. Similarly, many banks' technology portfolios are not aligned with the types of drivers which have been shown to beat long-term market total shareholder returns. For example, with "change the bank" spend (initiatives that aren't just about "run the bank" maintenance) being often significantly less than half a bank's tech spend, it is not surprising that business leaders don't see technology investments generate top-line growth or reduce expenses.

Our McKinsey colleagues recently published a book called "Rewired," which highlights the importance of going beyond the tech itself to address issues such as where to play, clarifying roles and responsibilities, breaking organizational silos or creating a compelling vision for investors and stakeholders to motivate people and drive change with urgency. One rule of thumb across successful institutions we have seen is that for every dollar invested in technology, it takes another dollar to be invested in this kind of strategic organizational, cultural and change management to be assured of the value.

Against this background, AI now looms ever larger for banks. But to capture any meaningful value from AI, the actions need to reach much further than just building sophisticated models. For example, at some institutions, even the process of validating machine learning or AI models can stretch to as long as two years. While there are often good reasons for this duration, in many cases relooking at these processes can compress the time taken while preserving the risk management (and sometimes even enhancing it). Similarly, many institutions have lots of pilots but little assurance about how they will scale those pilots.

Payments

JPMorgan, Commonwealth Bank combine AI tools to improve payments

The two financial institutions are testing a combination of machine learning and blockchain to catch errors and block fraud on international transactions.

By John Adams

June 25

Over the last 15 years, banks have seen many trends that held the promise to change their business, like lean, agile, robotic process automation, core platform modernization or the cloud. Consequently, many institutions are still on those journeys. And now they are faced with taking on AI. At this week's conference and from our work with clients, it is becoming clear how the deployment of AI holds the promise to change the odds in favor of banks (our colleagues estimate this at $200 to $340 billion in value from generative AI globally, and significantly more from advanced analytics adoption overall).

Yet as this new "wonder" technology takes hold, the big lesson of past and ongoing technology implementations needs to be kept in mind — namely that the impact of tech will need to be captured outside the CIO's office. To that end, we see three critical sets of questions for banking leaders as they head home from this week's gathering in Florida:

First, can you objectively identify areas where tech/AI can generate the most business value in your context (e.g., reducing risk, introducing new revenue, cutting cost)? 

Second, are you materially reallocating your spending toward those areas or are you being incremental and "peanut buttering" investments — this includes funding the required changes in enabling functions like data, or risk management and compliance?

Third, beyond the tech or AI model deliverable, do you have a change management formula in place that reaches "beyond the CIOs office" and benefits from your past lessons from other similar epochal programs? As our research last year found, those banks that foster integration between technical talent and business leaders are more likely to develop scalable AI solutions that create measurable value.

While we have seen many "digital-native banks," the world has yet to see an "AI-native bank." To get there, banks will need to internalize the great lesson of the past — that ironically, the secret to successfully deploying any technology is never just technology.