Page 29 - gs260301

Insights and Expertise

There is an argument to be made for a longer range view.
The emphasis has been on how AI can be deployed at
scale to improve financial flows, logistics, infrastructure Regulation alone will not
and coordination across entire systems. The payoff of that
approach is now becoming visible. Long-term planning resolve this. Nor will simply
compounds. Short-term optimization eventually hits con- building more powerful
straints.
models. The deeper issue
We often get told we have lost the art of cathedral think-
ing: the mindset that accepted a project might take genera- is what we are choosing to
tions to complete. Modern political cycles, typically four or
five years, reward visible short-term wins over structural optimize for.
investment, and the same logic can apply in technology.
An instant-gratification mindset is unlikely to produce re-
silient AI services, but a longer horizon might.
agement systems depend on continuity. A wave of failed
Regulation alone will not resolve this. Nor will simply vendors and inflated claims can make buyers more cau-
building more powerful models. The deeper issue is what tious, even toward solutions that warrant confidence.
we are choosing to optimize for. Longevity, resilience and
shared utility tend to deliver slower initial returns, but This is where a more collective approach becomes impor-
they also reduce the likelihood of sharp corrections. In fi- tant. Investing in resilient infrastructure, shared standards
nancial services, that trade-off is well understood. and interoperable systems may not generate headlines, but
it ensures that when the hype subsides, the foundations
Fintech already knows what durable AI looks like remain solid.

In payments and fintech, AI has been delivering value qui- Bringing it back to payments
etly for years. Fraud detection, transaction monitoring and
risk scoring systems are not speculative. They are embed- For payments, the takeaway is straightforward. Reinforc-
ded, measured and continuously improved. They do not ing the rails matters more than chasing novelty. Systems
promise transformation overnight. They promise fewer that reduce fraud, optimize routing and protect smaller
false positives, lower losses and more resilient networks. businesses create compounding benefits across the eco-
system. They also stand the test of time.
That is why the current fixation on generative AI and large
language models deserves caution rather than dismissal. Feathering individual nests through speculative AI de-
Some applications will prove useful. Many will not. Treat- ployments may deliver short-term gains. It does not create
ing all AI as if it were synonymous with LLMs distorts long-term stability. At scale, that approach increases the
investment decisions and boardroom priorities. likelihood of the very bubbles the industry claims to fear.

A widely cited figure suggests the majority of AI initia- A slowdown in AI investment does not have to be destruc-
tives fail to deliver their expected returns. That failure tive. Managed with discipline, it can provide clarity. It can
rate is not a technology problem. It is a problem of intent, distinguish durable infrastructure from disposable exper-
evaluation and discipline. When tools are adopted for vis- imentation, and long-term systems from short-term noise.
ibility rather than utility, disappointment is the predict-
able outcome. The AI bubble has not burst. But when momentum inevi-
tably cools, the organizations that endure will not be those
For fintech, the risk is not that AI ceases to function. It that chase attention. They will be those that build for lon-
is that infrastructure-level applications become collateral gevity, not simply for launch.
damage from overpromising elsewhere.

What a lull would expose Scott Dawson, head of sales and strategic partnerships at DECTA, is a
highly motivated and results oriented individual with 20 years of experi-
If the market moves into a slower phase rather than a full ence within the payments industry. Previously, he served as commercial
crash, the consequences will still be meaningful. Capital director at Neopay. He has also held fraud management positions at PSI
will tighten. Procurement cycles will extend. Experimen- Holdings and Neteller, before becoming senior fraud manager and then
tal deployments without a clear operational rationale will business development manager at ClickandBuy, which was acquired by
likely be cut first. That is not catastrophe; it is recalibra- Deutsche Telekom. DECTA provides end-to-end payment infrastructure,
tion. from acquiring to issuing and processing, but unlike other players in
the crowded payments marketplace the company offers bespoke-as-
The greater risk lies in allowing that recalibration to erode standard solutions aimed at making payments accessible to everyone.
trust in AI systems already performing essential work. Contact Scott via LinkedIn at linkedin.com/in/scott-dawson-uk.
Payments infrastructure depends on reliability. Risk man-
29