Only 19% say they have “good enough” metrics and KPIs for their current stage of AI deployment. need be�er metrics to measure the success of their AI implementa�ons 72% need be�er financial KPIs for AI-enabled opera�ons 25% need be�er non-financial KPIs for AI-enabled opera�ons 27% need both 21% Q. Which statement most closely matches how you feel about measuring the success of and financial return on AI implementations? The issue of inadequate evalua�on methods is further highlighted in the field of Genera�ve AI, specifically. According to the April 2024 Ar�ficial Intelligence Index Report from Stanford University's Ins�tute for Human-Centered Ar�ficial Intelligence: "Robust and standardized evalua�ons for LLM responsibility are seriously lacking." The report emphasizes that leading developers, such as OpenAI, Google, and Anthropic, primarily test their models against different AI benchmarks, which complicates efforts to systema�cally compare the risks and limita�ons of top GenAI solu�ons. This lack of standardiza�on in responsible AI repor�ng poses significant challenges for businesses looking to invest in and implement advanced AI solu�ons. Without a clear, industry-wide framework for assessing the effec�veness and responsibility of AI models, organiza�ons may struggle to make informed decisions about which technologies to adopt and how to measure their success. As a result, businesses may miss out on the poten�al benefits of AI. To address these challenges, it is crucial for industries to collaborate on developing robust, standardized evalua�on methods that enable businesses to assess the performance and responsibility of AI implementa�ons accurately. By establishing clear benchmarks and repor�ng standards, organiza�ons can make more informed decisions about their AI investments, ul�mately leading to more successful implementa�ons and greater overall business value. TCS AI for Business Study Key Findings Report 26