While AI and machine-learning (ML) projects are a high priority for many companies today, a report by Dimensional Research commissioned by Alegion has found that eight out of 10 organizations reported their AI and ML efforts have stalled. Moreover, 96% of companies have experienced problems with the data quality and labeling that is required to train AI and build model confidence.
The report, Artificial Intelligence and Machine Learning Projects Obstructed by Data Issues, was created using feedback from 227 participants, including data scientists and business stakeholders involved in AI and ML efforts.
"The single largest obstacle to implementing ML models into production is the volume and quality of the training data," remarked Nathaniel Gates, CEO and co-founder of Alegion, a training data platform for AI and ML initiatives. "This research reinforces our own experience, that data science teams new to building ROI-driven systems try to tackle training data preparation in house and get overwhelmed."
Most teams in the survey reported that they were underwhelmed because they had underestimated the difficulty of implementing the technologies, with 81% admitting the process of training AI with data was more difficult than they had anticipated. In an attempt to overcome the difficulties, 76% revealed they had attempted to label and annotate training data on their own. Some 71% of respondents, however, ultimately outsourced this task as well as other ML projects.
Dimensional Research also discovered that most organizations' AI efforts were relatively recent, with 70% answering that they had begun their first AI/ML investment in the last 24 months.