View moreThoughts from the CEO — Google Cloud Next 2025 Recap
Last week at Google Cloud Next 2025 in Las Vegas, I had the opportunity to sit on a panel titled Utilizing Artificial Intelligence for Good. We talked about ethics, privacy, deployment strategies, and organizational readiness — but one moment stood out above the rest.
A gentleman, an AI engineer from the World Bank, posed a great question near the end of the session and a group formed around me after the general session — they all had the same problem:
“We’ve been deploying AI models via APIs across various engineering teams, but we keep hearing from internal users that the results just aren’t helpful. What can we do to fix that?”
It was a sincere question — and one I think many institutions are quietly asking themselves. Frankly, I hadnt been presented with this question. I thought for a few moments and asked them a question.
“Are you using tuned, and models trained on specific use cases, or general AI?”
They all told me general models. My answer was simple:
“Get specific. Just like you have specialists in your organization, your AI needs to be fine-tuned and trained for specific purposes. General models don’t solve specialized problems.”
At BankSocial, that principle is central to everything we do. Across our systems — whether they’re member-facing workflows, employee productivity tools, or fraud detection engines — we don’t rely on a single model to do it all. For example, we are currently in the process of rolling out 19 uniquely trained models just for financial data alone.
And that brings me to something we deeply believe in:
Human-Powered AI ™
We don’t see AI as a replacement for people — we see it as a force multiplier for human intelligence. Just as people tend to “hallucinate” or get things wrong when faced with questions outside their expertise, AI models can do the same. But just like a human learns through training, trial, and error, so too can AI be shaped through intentional fine-tuning. At BankSocial, we believe AI should serve people, not replace them. That’s why we’ve built a cultural and operational foundation around what we call our Three Laws of AI Engagement — designed to ensure our models enhance human capability, not compete with it.
These aren’t just guidelines — they’re mandates that align AI strategy with real business outcomes.
1. AI Before Headcount
We’ve made it policy to explore AI solutions before adding new roles. For repeatable, high-volume tasks like reconciliations, risk modeling, or initial fraud screening — we deploy AI first. It allows us to scale smartly and maintain agility while keeping our teams focused on higher-order thinking.
2. AI as a Core Skill
In our organization, AI literacy is not optional. We’re training all team members to prompt, contextualize, and validate AI outputs. Whether you’re in product, ops, compliance, or service — knowing how to work with AI is now as essential as knowing how to use email. Google’s free training tools are helping accelerate this across the board.
3. Reward AI Adoption
We tie performance reviews to AI engagement. When teams use AI to improve accuracy, speed up reporting, detect fraud faster, or deliver better service — we recognize and reward that. Because innovation only takes root when it’s part of the culture.
This mindset — Human-Powered AI, backed by structure and intentionality — is what we believe the financial services industry needs more of.
As I told the World Bank team:
“If your AI isn’t working, it’s probably because it’s too generic. AI needs purpose. It needs tuning. And most importantly, it needs people.”
AI is not a silver bullet — it’s a highly specialized tool that must be guided, trained, and continuously refined. That’s what we’re doing at BankSocial. And that’s why it’s working.
From fraud detection to member engagement to internal operations — our Human-Powered AI approach is built for real-world results.
If you’re a credit union or bank leader trying to figure out how to make AI actually useful in your institution, let’s talk.
was originally published in BankSocial News on Medium, where people are continuing the conversation by highlighting and responding to this story.
