Opinion: AI needs rules and standards, or it may end up being more risky than profitable

The promise of profound new types of productivity and unimaginable intelligence is significant, but with that potential also comes unprecedented fear. 

Leaders face uncertainty about how to adapt to the technology, optimize it and ultimately harness its power. I am a big believer in unleashing AI-powered innovation and putting it into the hands of as many people as possible to unlock capabilities we’ve never seen or imagined before. But just as quickly, companies need guidance on how to experiment, pilot and adopt AI ethically and securely. 

Here’s an AI governance roadmap for leaders to guide their efforts in 2024 and beyond:

1. Trusted data: In my world, so-called unstructured data requires refinement, cleaning and sorting to capture nuggets of business intelligence. But AI is only as good as the data that’s powering it, and so the first step of strong AI governance is to ensure you can trust the data itself. 

According to a recent survey conducted by the Eckerson Group, 46% of data leaders said that their organization lacks adequate data quality and data governance controls. Even then, most business leaders are not striving for adequate – they are striving for excellence.

What does that look like? High-quality data is accurate, complete, consistent, timely, valid and unique. For example, in the healthcare industry, it’s essential to have complete, correct and unique patient records with no duplication to ensure proper treatment, monitoring and billing. Just one mistake in managing that data can have costly — even devastating — consequences, and that is before AI or large language models (LLMs) for generative AI are built on top of it.  

The challenge, of course, is that most organizations still have highly fragmented data and lack a comprehensive governance framework. The natural starting point is assessing the current state of your data — what you have, where it lives, how it moves and how it’s protected to diagnose any quality issues. Then apply rules to manage and monitor it to ensure quality is maintained over time. 

2. Modern governance: Data governance is not a new concept, but it is increasingly important amid the proliferation of AI and generative AI applications, stakeholder and shareholder demand, as well as global data regulations.

Many leaders now have outlined data governance practices that cover the people, processes and technologies across their organizations to create common rules and guidelines for collecting, storing and using data. Most of these frameworks are risk and compliance based — and for good reason. Ethically securing, protecting and using data is an essential part of any successful data strategy to align with current and future data and AI regulation. U.S. President Joe Biden’s recent executive order on AI, for example, will increase companies’ reporting standards and the government’s oversight of AI, while the EU is poised to announce AI regulation and enforcement practices in short order. 

Still, companies’ traditional approaches to governance will be found wanting when it comes to ensuring safe and effective data management for AI and generative AI. Why? For the most part, they weren’t designed for the scale or democratization of data-use that businesses now require. The sheer volume, variety and velocity of data is too much for human management alone. 

Modern data governance builds on those table-stakes features of security and compliance to incorporate automation, adaptability and agility. It both considers and automates data integration, cataloging, management and observability to increase productivity and outcomes — in effect becoming an exercise in competitive advantage, not just compliance. 

3. Getting started: AI will be a top priority as leaders finalize their organizations’ 2024 strategies. This is driven in part by democratization and growing access to generative AI. In fact, McKinsey reports that 40% of organizations will invest more in AI overall because of the advances in generative AI. But the smartest companies will also make AI governance a priority in order to maximize value and minimize risk. 

To begin, assess your data’s current quality and any underlying issues or disconnects in your data management processes. Form a collaborative team to develop and oversee your governance practices to ensure ongoing accuracy and reliability of your data and AI-powered insights. 

Finally, keep iterating. Just as feedback is essential to product development and improvement, it’s equally important to AI models’ performance. Your approach should be flexible to consider new innovations and ways of using the technology and, more importantly, new security threats that are emerging each day. 

While we’ve never seen anything like AI, history is filled with countless innovations and opportunities that caused fear and required guardrails to be built, from the gold rush and transportation boom in the mid-1800s to telecom and pharmaceutical developments in the 20^th century. 

History has shown time and again that through the public and private sectors, we can and will rise to meet the challenge. AI will be the same if we put the technology in the hands of as many people as possible and collaborate to secure our future with it. 

Amit Walia is the CEO of Informatica, an enterprise cloud data management platform company. 

More: AI needs a strong code of ethics to keep its dark side from overtaking us

Also read: AI may not replace your job, but it’s going to remake your work. One thing must not change.