AI Hype CRASHES—CEOs Spill Truth

Digital,Microcircuit,Background,Technology,Motherboard

AI hype promises instant job-killing automation, but top CEOs admit it’s far harder than advertised—exposing overblown tech dreams that could waste billions in American investments.

Story Snapshot

Databricks and Glean CEOs reveal AI automation demands massive engineering and human oversight, debunking quick-deployment myths.
95% failure rate in projects signals healthy experimentation, not instant success, despite billions in recent funding.
Human-AI “super teams” emerge as reality, protecting American jobs from premature displacement amid reskilling needs for 31% of workforce.
Under President Trump’s pro-innovation policies, realistic AI scaling bolsters U.S. leadership without globalist overpromises.

CEOs Challenge AI Hype on Podcast

Ali Ghodsi of Databricks and Arvind Jain of Glean spoke on the December 23, 2025, “Bg2 Pod” episode. They stated AI automation requires extensive engineering effort beyond simple agent deployment. Ghodsi emphasized AI as an “engineering art” needing evaluation teams and human oversight. Jain shared Glean’s internal failures, like unfulfilled employee priority automation and custom model tuning. Both CEOs framed 95% project failure rates as positive signs of innovation. Their candor counters promotional narratives pushing rapid adoption.

Massive Funding Amid Deployment Realities

Glean raised $150 million in September 2025 at a $7.2 billion valuation. Databricks secured over $4 billion in early December 2025, reaching $134 billion valuation. These funds highlight sector growth, yet CEOs warn business leaders overestimate ease. Ghodsi noted agents do not “just work” upon release. Jain said successes take longer than expected. Firms now favor foundation models over custom ones due to complexities. This aligns with President Trump’s focus on practical American tech leadership.

AI agents advanced from one-hour autonomy in May 2025 to over 30 hours by September, per BCG reports. Task lengths double every 3-7 months, but reasoning and coordination gaps persist. Legacy system integration and data privacy barriers slow progress. Such realities temper hype, ensuring investments yield real gains rather than wasteful flops.

Human Oversight Essential for Progress

Ghodsi predicted in June 2025 that humans must supervise AI agents long-term. Jain views high failures as experimentation fuel. This human-in-the-loop approach forms “super teams,” vital as 31% of workforces need reskilling in three years. Surveys show executives optimistic—51% expect customer gains by 2026—yet deployments lag. UPS succeeded in route optimization, contrasting Glean’s setbacks. Common sense demands realism to protect jobs and efficiency.

President Trump’s policies attract over $1 trillion in AI investments, solidifying U.S. dominance. Deregulation and innovation incentives enable sustainable scaling without Biden-era overregulation. Job displacement risks—400-800 million globally by 2030—underscore need for measured adoption. Enterprises delay full automation, prioritizing workflow redesigns for 70% faster decisions.