In this era of AI innovation, organizations are spending tens of billions of dollars on AI development. However, for all the money invested in capabilities, there has not been a corresponding investment in AI governance.
Some companies may take the position that when world governments loosen AI regulations, it will be an opportune time to fight AI programs into governance structures that can address complex topics like privacy, transparency, accountability, and justice. In the meantime, businesses can focus solely on AI performance.
The wheels of adjustment are in motion. However, regulations move at bureaucratic speed, and AI innovation only accelerates AI has been deployed at scale, and we are rapidly approaching the point after AI capabilities will exceed effective regulations, putting the responsibility for self-regulation squarely in the hands of business leaders.
The solution to this conundrum is for organizations to find a balance between existing rules and their own rules. Several companies are rising to the challenge of responsible AI: Microsoft has an Office for Responsible Use of AI, Walmart has a Digital Citizenship team, and Salesforce has an Office for the Ethical and Humane Use of Technology. However, more organizations need to quickly embrace the new era of AI.
Business value in self-regulation
Government agencies cannot scrutinize every company, understand at a technical level what AI programs are emerging, forecast potential problems that may occur, and then quickly create rules to prevent problems before they occur. That’s an unattainable regulatory scenario—and no business wants that in any case. However, each company has a clear view of its own AI efforts, putting it in the best position to address AI problems as they are identified.
When government regulations are enforced by fines and litigation, the consequences of failing to regulate ourselves have the potential to be even more impactful.
Imagine an AI tool embedded in a retail setting that uses CCTV feeds, customer data, real-time behavioral analysis, and other data to predict what shoppers might buy if employees use certain sales techniques. AI also forms customer personas that are saved and updated for targeted advertising campaigns. The AI tool itself is purchased from a third-party vendor and is one of dozens of AIs deployed throughout the retailer’s operations.
Emerging regulations could dictate how customer data is stored and transferred, whether consent is required before data is collected, and whether tools are fair in their predictions. These considerations are valid, but not comprehensive from a business perspective. For example, have AI vendors and their tools been vetted for security gaps that could jeopardize the company’s connected technology? Do staff have the necessary training and documented responsibilities to use the tool properly? Do customers know that AI is being used to build detailed personas that are stored elsewhere? Does he need to know?
The answers to these questions can significantly impact the company in terms of security, efficiency, ROI on technology investments, and brand reputation, among others. This hypothetical case illustrates how failing to self-manage AI programs exposes organizations to a wide range of potential problems – many of which may fall outside the purview of government regulations. The best path forward with AI is being shaped by governments.
Government for confidence in AI
No two companies and AI use cases are the same, and in the era of self-regulation, companies are asked to assess whether the tools they use can be deployed safely, ethically, and in line with company values and existing or tangential rules. In short, businesses need to know that AI can reliable.
Trust as a lens for governance not only raises common AI issues, such as potential discrimination and threats to the security of personal data. As I discuss in my book, Trusted AItrust also applies to things like reliability over time, transparency for all stakeholders, and accountability created throughout the AI lifecycle.
Not all of these factors are relevant for every organization. AI that automates trade reconciliation may not pose a threat of discrimination, but the security of the underlying model and data is critical. On the other hand, data security is somewhat less for predictive AI used to anticipate food and housing security, but unfairness and discrimination are priority considerations for tools that rely on historical data with potential for latent bias.
Effective self-regulation in AI requires a whole-life-cycle approach, where attention to trust, ethics, and outcomes is embedded in each project phase. The process needs to be changed to set a clear direction point for decision making. Employees must be educated and trained to contribute to AI governance, with a solid understanding of these tools, their impact, and the individual responsibilities of employees in their lifecycle. And the technology ecosystem of edge devices, cloud platforms, sensors, and other devices must all be aligned to promote the quality of trust that is most important in a given deployment.
Self-regulation fills the gap between innovation and government-made rules. It not only sets the company on a path to meet any regulations that arise in the future, but also provides significant company value by maximizing investment and minimizing negative outcomes.
For all that we’ve invested in building AI capabilities, we also need to invest in how to manage and use these tools to their potential in ways that we can trust – and we shouldn’t wait for governments to tell us how.
Beena Ammananth is the executive director of Global Deloitte AI Institute.
Opinions expressed in Fortune.com comment pieces are solely the opinions of the authors and do not necessarily reflect their opinions and beliefs. fortune.
Another must read comment published by fortune:
Learn how to navigate and strengthen trust in your business with The Trust Factor, a weekly newsletter that examines what leaders need to succeed. Log in here.