The swift adoption of Generative AI over the past year has transformed the AI landscape, stressing the need for Responsible AI programs to keep pace. Over half (53%) of organizations solely use third-party AI tools without any internal AI, leading to 55% of all AI-related failures, according to recent research by MIT Sloan Management Review (MIT SMR) and Boston Consulting Group (BCG).
The report, "Building Robust RAI Programs as Third-Party AI Tools Proliferate," was based on a survey of 1,240 participants, representing organizations with a minimum of $100 million annual revenues, from 59 industries across 87 countries.
The report found that 78% of surveyed organizations heavily rely on third-party AI, exposing them to numerous risks, including reputational damage, loss of customer trust, financial loss, regulatory penalties, compliance issues, and litigation. One-fifth of these organizations do not evaluate these risks. The report encourages using multiple approaches to evaluate third-party tools for risk mitigation. Organizations utilizing seven different methods are more likely to identify lapses than those using only three methods (51% vs. 24%).
The regulatory landscape is rapidly changing with numerous AI-specific regulations. About 51% of organizations surveyed are subject to non-AI-specific regulations affecting their use of AI. These organizations have 13% more Responsible AI Leaders and report fewer AI failures (32% vs. 38%) than those not under similar regulatory pressures.
The report offers five recommendations for organizations facing the rapid adoption of Generative AI and its associated risks: (1) advance Responsible AI programs, (2) effectively evaluate third-party tools, (3) prepare for emerging regulations, (4) involve CEOs in Responsible AI initiatives for maximum success, and (5) increase investment.
