What It Takes to Make AI Safe and Effective


– By Avivah Litan


AI TRiSM capabilities ensure model reliability, trustworthiness, security and privacy.


Organizations that apply rigorous artificial intelligence trust, risk and security management (AI TRiSM) move more valuable AI models into production.


In today’s digitalized world, a rigorous approach to AI TRiSM is needed to build safeguards into the AI models and strategies. Gartner defines AI TRiSM as a framework that supports AI model governance, trustworthiness, fairness, reliability, robustness, efficacy and privacy. It includes solutions, techniques and processes for model interpretability and explainability, privacy, model operations and adversarial attack resistance for its customers and the organization.


IT leaders must spend time and resources on supporting AI TRiSM. Those who do will achieve improved AI outcomes in terms of adoption, business goals and both internal and external user acceptance. AI threats and compromises (malicious or benign) are continuous and constantly evolving, so AI TRiSM must be a continuous effort, not a one-off exercise.


Why AI TRiSM Is a Trending Technology


Gartner expects that by 2026, organizations that operationalize AI transparency, trust and security will see their AI models achieve a 50% result improvement in terms of adoption, business goals and user acceptance. Gartner also predicts that by 2028, AI-driven machines will account for 20% of the global workforce and 40% of all economic productivity.


But Gartner survey results indicate that organizations have also deployed hundreds or thousands of AI models that IT leaders can’t explain or interpret.


Organizations that don’t manage AI risk are much more likely to experience negative AI outcomes and breaches. Models won’t perform as intended, and there will be security and privacy failures, financial and reputational loss, and harm to individuals. AI that is carried out wrongly can also cause organizations to make poor business decisions.


AI TRiSM Implications and Operations


AI regulations are increasing, but even before protections are mandated, it is important to implement practices that ensure trust, transparency and consumer protection. IT leaders need to apply new AI TRiSM capabilities to ensure model reliability, trustworthiness, privacy and security.


Don’t wait until models are in production to apply AI TRiSM. It just opens the process to potential risks. IT leaders should familiarize themselves with forms of compromise and use the AI TRiSM solution set so they can properly protect AI.


AI TRiSM requires a cross-functional team to work together. This includes staff from the legal, compliance, security, IT and data analytics teams. Set up a dedicated team if possible, or a task force if not, to gain the best results. Ensure appropriate business representation for each AI project.


Benefits include improving the business outcomes their organization derives from its use of AI, rather than simply complying with regulations.


Gartner analysts will further discuss this topic at the Gartner Data & Analytics Summit 2023, taking place from May 8-9, Mumbai, India.



(The author is Avivah Litan, VP Analyst, at Gartner, and the views expressed in this article are his own)



Leave a Response