AI CEOs say automation with AI is harder than it looks (www.businessinsider.com)

In a recent discussion on the "Bg2 Pod," Ali Ghodsi, CEO of Databricks, and Arvind Jain, CEO of Glean, warned that automating work with AI is far more complex than corporate leaders may realize. Jain recounted his challenges in trying to automate internal workflows at Glean, highlighting that initial assumptions about AI's capabilities can lead to disappointment. Despite having comprehensive company context, efforts to use AI for tasks like prioritizing employee goals fell short. Both executives emphasized that the reality of implementing AI often involves a significant investment of time and resources in engineering and experimentation, which can sometimes lead to failures—viewed as a necessary part of innovation. This insight is crucial for the AI/ML community as it underscores the need for realistic expectations regarding AI deployment in organizations. Ghodsi pointed out that successfully integrating AI is akin to an "engineering art," demanding careful planning and human oversight. This sentiment was echoed by industry leader Yoshua Bengio, who noted the increasing importance of human qualities in a landscape where machines perform routine tasks. As AI technologies evolve, balancing automation with human input will remain essential to drive meaningful advancements and avoid the pitfalls of overreliance on unproven AI solutions.