Databricks CEO says AGI is already here — and Silicon Valley just keeps moving the goalposts (www.businessinsider.com)

Databricks CEO Ali Ghodsi told Goldman Sachs’ Communicopia + Technology Conference that, by the standard definition researchers used a decade ago, AI chatbots already qualify as artificial general intelligence (AGI). He argued that Silicon Valley shifted the target to “superintelligence” — systems far smarter than humans — after AGI-like capabilities arrived, and that this elevation is distracting. Ghodsi, who holds a CS doctorate, said companies don’t need superintelligence to automate workflows or build intelligent agents; instead the work ahead is practical engineering and productization (“the boring work”). Databricks’ recent $1 billion raise and >$100B valuation underscore this enterprise focus. Technically, Ghodsi pointed to diminishing returns from scaling pre‑trained giant models: scaling laws that drove recent advances have slowed and successors such as GPT‑5 or Claude 4 aren’t delivering massive leaps. His view implies a pivot from chasing ever‑larger foundation models toward systems integration, agent orchestration, and incremental value extraction. The industry is divided: Microsoft’s AI chief calls superintelligence an “anti‑goal,” while OpenAI and DeepMind leadership explicitly aim for superintelligence within years. The debate shifts the question from “when will AGI arrive?” to “what practical capabilities and governance frameworks should we deploy now?”