Data Realities Drag on AI As organizations aggressively invest to apply AI to an expanding set of objectives, “a kink is emerging in organizations’ project pipelines,” suggest new findings from S&P Global. While more initiatives are funneled toward AI project teams, S&P Global analysts noted a buildup of initiatives that have been only partially deployed. On average, organizations surveyed have more projects classified as being in production with a limited deployment than ones with scaled-up capabilities. In the average organization, shows S&P Global data, 51 percent of AI projects are in production but not being delivered at scale. The crux of the problem, said the research firm, appears to be data quality and availability, with legacy data architectures causing this pipeline stoppage in many organizations. All the while, the constant chasing of new initiatives means many organizations fail to maximize the value of their existing investments, the research firm warned. “AI projects risk stalling in a limited deployment purgatory, costing a company money, time and resources, while not seeing desired levels of use,” said the S&P Global study, commissioned by AI company Weka. “Initiatives are becoming snagged on data siloes, poor data quality and ineffective data and model pipelines.” Data quality is the most frequently identified challenge as organizations move their projects from pilots to production, identified by 42 percent of organizations as among their top three barriers. That placed data quality as an even more significant issue than skill shortages (32 percent) and budget limitations (31 percent). Organizations in media and By Martin Vilaboy 10 THE CHANNEL MANAGER’S PLAYBOOK
RkJQdWJsaXNoZXIy NTg4Njc=