AI CEOs Admit Their Multi-Billion Dollar Bots Are Still On Training Wheels
KEY POINTS
- •Databricks and Glean CEOs Ali Ghodsi and Arvind Jain spoke on the 'Bg2 Pod' about AI automation challenges, urging more realistic expectations.
- •Jain revealed his startup's $150 million funding and failed attempts to automate internal workflows and customize AI models thwarted progress.
- •Ghodsi highlighted that despite Databricks' $134 billion valuation and recent $4 billion funding, human oversight remains essential for AI deployment.
In a revelation that might surprise precisely nobody except overly optimistic CEOs, Databricks and Glean bosses Ali Ghodsi and Arvind Jain confessed AI can’t just 'magically' automate work despite their companies being valued at $134 billion and $7.2 billion respectively. Jain, who oversees an AI startup flushed with $150 million from September fundraising, admitted attempts to automate internal workflow and build bespoke models failed spectacularly, pushing them back to simpler off-the-shelf solutions. Ghodsi, whose company just bagged $4 billion last week, calls AI integration an 'engineering art' requiring 'human supervisors' for all those messy machine decisions. According to Jain, a 95% project failure rate is not a bug, but a feature—because apparently trying and failing is the new success mantra at multi-billion dollar AI firms. Humans still firmly in the loop, apparently to click 'OK' when the ‘magic’ inevitably misfires.
Share the Story
Source: Businessinsider | Published: 12/24/2025 | Author: Lee Chong Ming