Is AGI already here, and are we missing the point? That's the provocative question Databricks CEO Ali Ghodsi is posing to the tech world. He believes the industry has already achieved Artificial General Intelligence (AGI) – AI that can reason like a human – but is now obsessed with an even more ambitious goal: superintelligence. But here's where it gets controversial... Ghodsi's argument, made at Goldman Sachs' Communicopia + Technology Conference, is that we've been moving the goalposts. He claims that the definition of AGI, as understood a decade ago, is already met by current AI chatbots.
He suggests that the focus on superintelligence, AI that surpasses human intellect, is "misdirected." Ghodsi, with a Ph.D. in computer science, argues that superintelligence isn't even close to being achieved with current technology and isn't what companies truly need. Instead, he believes AGI is sufficient to automate tasks and build effective AI agents. And this is the part most people miss... He emphasizes the need to focus on the "boring work" of implementing and refining these existing AGI capabilities. Databricks, a company valued at over $100 billion after raising $1 billion in September, is clearly invested in this vision.
Ghodsi also points out that the era of massive leaps in AI model performance may be slowing down. He notes that the scaling laws that fueled recent AI progress have "come to a stop," with newer systems like OpenAI's GPT-5 and Anthropic's Claude 4 not delivering dramatic improvements. In his words, "It's getting harder and harder to get value out of the next pre-trained giant model."
But what about the future of AI? The debate surrounding superintelligence is heating up. Microsoft AI CEO Mustafa Suleyman has even called superintelligence an "anti-goal," fearing its potential for misalignment with human values. He advocates for "humanist superintelligence," which prioritizes human interests.
However, other tech leaders are pushing forward with the pursuit of superintelligence. OpenAI CEO Sam Altman aims to achieve it, believing it could dramatically accelerate scientific progress and prosperity. He predicts its arrival by 2030. Google DeepMind's cofounder, Demis Hassabis, shares a similar timeline, suggesting AGI could arrive within the next 5-10 years, with AI systems deeply integrated into everyday life.
What do you think? Is AGI already here, and are we getting distracted by the pursuit of superintelligence? Share your thoughts in the comments below! Do you agree with Ghodsi's assessment, or do you believe superintelligence is the ultimate goal?
