A recent survey conducted by machine learning firm Tecton reveals a surge of interest in applied AI/ML among organizations. Despite facing challenges related to data accuracy, production pipelines, and demonstrating ROI, the survey respondents revealed their commitment to bolstering their applied ML capabilities.
Surge in diverse ML use cases
The survey shows a rising tide of businesses harnessing the power of applied ML across various functions. From customer analytics to personalized recommendations and fraud detection, organizations are exploring the versatility of applied ML. The increasing adoption marks a shift towards more targeted and efficient business practices achieved through machine learning.
Despite the enthusiasm for applied ML, the journey is not without hurdles. The survey reveals three main challenges faced by companies: generating accurate training data, building production data pipelines, and demonstrating business ROI. These obstacles highlight the complexities and intricacies of implementing applied ML and provide a roadmap for areas that require attention and improvement.
Commitment to enhancing ML capabilities
In spite of the challenges, organizations remain committed to harnessing ML. They are focusing on improving model deployment time, adopting real-time machine learning, and implementing centralized ML platforms. These initiatives underscore the idea of cross-team collaboration and organizational scalability, essential for successful ML integration and use.
ML tops priority list for organizations
The survey reveals that a significant number of organizations are prioritizing applied ML. For 23.8% of respondents, it's the number one initiative, while for 60.1%, it's among the top three. This indicates the increasing importance and reliance on machine learning in the business sector, reflecting a promising future for applied ML.
Top challenges in model deployment
According to the survey, generating accurate training data, building production data pipelines, and demonstrating business ROI are the top challenges when deploying new models to production. These findings provide further insights into the specific roadblocks in the deployment stage, emphasizing the need for companies to focus on these areas for successful ML implementation.
The growing interest in real-time ML models is also evident. The survey found that 68.3% of the respondents already have at least one real-time ML model in production. Meanwhile, 14.7% have more than ten. This trend points to the increasing need for swift, real-time decision-making capabilities in businesses, facilitated by ML models.