AI Leaders Ground the AGI Debate in Reality

Tech leaders continue to debate whether artificial general intelligence (AGI) is just around the corner—or if it remains a distant goal. While some CEOs predict AI models will soon surpass human intelligence, a growing group of experts are skeptical, arguing that today’s large language models (LLMs) lack the creativity required for true AGI.

Anthropic CEO Dario Amodei has claimed that by 2026, AI could be “smarter than a Nobel Prize winner across most relevant fields,” while OpenAI’s Sam Altman has suggested that “superintelligent” AI is within reach. These leaders see advanced AI as a transformative force that will drive scientific breakthroughs.

However, AI experts like Hugging Face co-founder Thomas Wolf disagree. In a recent essay, Wolf called such predictions “wishful thinking at best.” He argues that AI excels at answering known questions but struggles with formulating entirely new ones—an ability crucial for groundbreaking discoveries.

Google DeepMind CEO Demis Hassabis has also expressed doubts, reportedly telling staff that AGI could be a decade away due to AI’s current limitations. Similarly, Meta’s Chief AI Scientist Yann LeCun has dismissed the idea that LLMs alone could achieve AGI, calling it “nonsense” and advocating for new AI architectures.

Kenneth Stanley, a former OpenAI researcher now at Lila Sciences, is working on AI creativity, a concept known as "open-endedness." He believes AI must go beyond reasoning and develop the ability to generate original ideas. “Reasoning models are goal-driven, but creativity requires divergence,” he told TechCrunch.

While AGI remains an uncertain goal, skeptics like Wolf, LeCun, and Stanley are pushing for a more grounded conversation—one that moves beyond hype to address the real challenges standing in the way of truly intelligent AI.




News Source: TechCrunch