Trust in Generative AI Surges, But Companies Lag in Safeguarding

Introduction

A compelling report from SAS and IDC has shed light on a prominent disparity in how businesses across the globe perceive and utilize Generative AI compared to traditional forms of artificial intelligence (AI). The 'IDC Data and AI Impact Report: The Trust Imperative' reveals that executives exhibit more confidence in Generative AI, despite a troubling lack of security measures in place.

Findings of the Report

Conducted among nearly 2,400 IT and business professionals worldwide, the study finds that a staggering 78% of respondents claim to fully trust AI systems. Yet, in a stark contradiction, merely 40% have invested in frameworks such as AI governance, ethical guidelines, and explainability protocols to back this trust. This discrepancy highlights a significant gap between perception and reality.

Companies prioritizing Trustworthy AI reap significant advantages, with a 60% higher likelihood of AI projects being profitable compared to those that neglect such considerations. Interestingly, those least willing to invest in creating a robust AI framework show an excessively optimistic perception of Generative AI, considering it three times more trustworthy than established AI technologies like Machine Learning.

Trust in Emerging AI Technologies

The report indicates a fundamental trend of fostering confidence in newer AI technologies. Nearly half (48%) of respondents assert full trust in Generative AI, while one-third exhibits confidence in Agentic AI. In sharp contrast, less than 20% express comfort with traditional AI technologies. However, concerns persist even around applications like ChatGPT, with respondents indicating anxiety surrounding data privacy (62%), transparency (57%), and ethical usage (56%).

The Current Landscape of AI Adoption

As businesses continue to rapidly adopt AI technologies, especially Generative AI, the associated risks and ethical dilemmas are also on the rise. The study noted that the awareness of Generative AI now eclipses traditional AI methodologies, garnering 81% recognition compared to 66% for the latter. Alarmingly, organizations have not made sufficient efforts to mitigate these risks through secure infrastructural investments. The study illustrates a lack of focus on AI Governance, with only 2% of respondents calling it a top priority, and fewer than 10% developing policies for responsible AI deployment.

Challenges in Data Management

The successful implementation of AI projects hinges heavily on effective data management and quality. The more autonomous AI systems become and the deeper they integrate into critical processes, the more crucial it is to have sophisticated data strategies. The IDC study identifies three major hurdles: poor data infrastructure, inadequate governance, and a deficiency in AI expertise. Nearly half of the companies surveyed cited decentralized data sources or non-optimized cloud environments as primary barriers to successful AI deployment. Other challenges included the absence of robust data governance processes (44%) and a decline in AI specialists within organizations (41%).

Additionally, 58% of companies reported that accessing relevant data sources constitutes the most significant barrier to their AI projects, alongside issues related to data privacy and compliance (49%) and data quality (46%).

Conclusion

As Bryan Harris, Chief Technology Officer at SAS, aptly points out, establishing trust in AI is paramount for the benefit of society, businesses, and employees alike. To cultivate this trust, AI providers must enhance their implementation success rates, individuals should remain critical of AI results, and executives need to equip their teams with the tools and expertise necessary for responsible AI usage. The challenges are many but so too are the opportunities for those who choose to prioritize trust and governance in their AI initiatives.

For comprehensive insights, the full report is available for download from the SAS website.