Thinker vs. Tinker: 76% of AI researchers doubt scaling alone will achieve AGI, despite Big Tech buildout

AI researchers overwhelmingly reject the tech industry’s scaling strategy for achieving artificial general intelligence (AGI), with 76 percent believing that simply throwing more computing power at existing models is unlikely to succeed. This skepticism comes as companies continue pouring billions into AI infrastructure, highlighting a critical disconnect between research consensus and industry investment strategies that could reshape the future direction of AI development.

The big picture: A new survey of 475 AI researchers by the Association for the Advancement of Artificial Intelligence reveals that the dominant industry approach of scaling up current models is widely considered insufficient for reaching human-level AI.

• 76 percent of respondents said it was “unlikely” or “very unlikely” that scaling current approaches would lead to artificial general intelligence.
• Stuart Russel, a computer scientist at UC Berkeley who helped organize the report, noted that “about a year ago, it started to become obvious to everyone that the benefits of scaling in the conventional sense had plateaued.”

Behind the numbers: The AI industry continues massive investment in scaling despite researcher skepticism, with venture capital funding for generative AI reaching over $56 billion in 2024.

• Microsoft alone has committed to spending $80 billion on AI infrastructure in 2025.
• These investments primarily fund massive data centers required to train and run generative AI models.

Why this matters: The disconnect between researcher perspectives and industry investment strategies suggests billions may be directed toward approaches with diminishing returns instead of more promising paths to advanced AI capabilities.

What researchers believe instead: Many AI experts point to the need for fundamental innovations rather than simply larger models with more parameters.

• Researchers suggest the field needs new architectures that can perform causal reasoning and better understand the world around them.
• The survey indicates greater focus is needed on developing AI that can grasp human intentions and goals.

Reading between the lines: The industry’s fixation on scaling may reflect short-term business priorities rather than the most scientifically sound path to advanced AI.

• Current large language models show impressive abilities in text generation but still struggle with basic reasoning tasks that humans find intuitive.
• The plateau in capabilities despite increasing model size suggests fundamental limitations in current approaches.

The counterpoint: Some researchers and companies remain committed to the scaling approach, arguing that current models haven’t yet reached their full potential.

• Proponents of scaling suggest that even larger models with more advanced training methods could still yield significant breakthroughs.
• The debate highlights a fundamental tension in AI development between engineering-focused scaling and science-focused understanding.

Looking ahead: As the limitations of current approaches become more apparent, the AI research community may shift toward more diverse and innovative methods for advancing AI capabilities.

• This could include hybrid approaches that combine aspects of current large language models with new architectures for reasoning and understanding.
• The survey results may influence future funding priorities and research directions within the AI community.