Conclusion About Scle Methodology Scale is fueling the generative AI revolu- This survey was conducted online within the United tion. Built on a foundation of high-quality States by Scale AI from February 20, 2024, to March 29, Whether you are building or applying AI, data and expert insight, Scale powers the 2024. We received 2,302 responses from ML prac- world’s most advanced models. Our years titioners (e.g., ML engineers, data scientists, devel- of deep partnership with every major opment operations, etc.) and leaders involved with model optimization and evaluation is key model builder enables our platform to AI in their companies. Participants who reported no empower any organization to apply and involvement in AI or ML projects were excluded from to unlock performance and ROI. evaluate AI. the dataset, resulting in a final sample size of 1800 respondents. scale.com The pace of innovation for generative AI continues A quarter of the respondents identified themselves as to accelerate. While the 2023 AI Readiness Report belonging to the Software and Internet/Telecommu- focused on how enterprises could adopt AI, this year’s nications industry (28%), with the Financial Services/ report examined challenges and best practices to Insurance Industry following closely behind at 15%. apply, build, and evaluate AI. The two most significant Business Services accounted for 7%, while the Gov- trends to emerge in our analysis are: ernment and Defense Industry represented 4% of the respondents. Among these industries, a majority of 1. The growing need for model eval- respondents specified their employment within the uation frameworks and private Information Technology department (33%). benchmarks. In terms of seniority within their organizations, nearly 2. The continued challenges a quarter of respondents (24%) identified themselves of optimizing models for as Team Leads, 22% as department heads, and 5% as specific use cases without owners. Sixty-six percent (66%) of respondents report sufficient tooling for involvement in AI model application and customization data preparation, model (applying AI), while 34% are directly engaged in de- training, and deploy- veloping foundational generative AI models (building ment. AI). Consequently, a significant portion of respondents (46%) represent organizations at an advanced stage of At Scale, our mission is to AI/ML adoption, with one to multiple models deployed accelerate the develop- to production and undergoing regular retraining. ment of AI applications. Approximately 26% are in the process of The Scale Zeitgeist: AI developing their inaugural model, while Readiness Report supports 23% are in the phase of evaluating that mission. We will potential use cases, underscoring continue to shed light on the significance and enthu- the latest trends, challenges, siasm for AI/ML project and what it really takes to development. build, apply, and evaluate AI. 46 47