AI has already run out of training data — but there's more waiting to be unlocked, Goldman's data chief says

Goldman Sachs’ chief data officer warns that the AI industry has reached "peak data," as the pool of high-quality public internet information is effectively exhausted. This scarcity threatens to slow the rapid development of new models, forcing a pivot away from scraping the open web toward utilizing synthetic data or proprietary corporate datasets. The shift highlights a critical dependency on private, enterprise-specific information rather than public knowledge. The reliance on machine-generated training material introduces significant risks, such as the proliferation of low-quality "AI slop" and potential creative plateaus. If models increasingly train on their own outputs, they may struggle to incorporate fresh human insights, raising philosophical concerns about the trajectory of artificial intelligence. This dynamic suggests that future advancements may depend less on raw data volume and more on the quality and context of available information. This issue is vital to open_data because it underscores the diminishing value of publicly available datasets for next-generation AI training. As public data becomes saturated, the strategic importance of high-quality, structured, and accessible enterprise data grows. It highlights a shift where the scarcity of useful open data drives interest in proprietary reserves, emphasizing that data usability and contextual relevance are now more critical than sheer quantity for sustaining AI innovation.

Source: businessinsider.com
Published on 2025-10-03

AI has already run out of training data — but there's more waiting to be unlocked, Goldman's data chief says

Related news