One of the most influential figures in AI has made a significant statement about the industry’s direction. OpenAI co-founder and Safe Superintelligence Inc. (SSI) CEO Ilya Sutskever declared in an interview with Dwarkesh Patel that “the age of scaling is over, and we’ve entered the age of research.”
The End of the “Age of Scaling”
According to Sutskever, 2020 to 2025 was the “age of scaling.” During this period, AI companies achieved rapid progress by investing in larger models, more data, and more compute.
However, that era is coming to an end. “The period when big data and compute almost guaranteed progress is giving way to the ‘age of research,’ where new fundamental ideas are needed,” Sutskever explained.
Limitations of Current Models
The most notable observation from the interview concerns the generalization capabilities of current AI models.
“The thing which I think is the most fundamental is that these models somehow just generalize dramatically worse than people,” Sutskever stated. Humans can learn efficiently from limited data and adapt to new situations, but current LLMs cannot achieve this.
This problem cannot be solved simply by adding more data or compute. New research approaches are required.
The State of Reinforcement Learning
Interestingly, Sutskever also commented on the role of reinforcement learning (RL) in current AI development.
“Based on various things some people say on Twitter, it appears that companies are now spending more compute on RL than on pre-training,” he noted. However, he also pointed out that RL provides “a relatively small amount of learning” for the compute it uses.
SSI’s Strategy
SSI, led by Sutskever, has raised $3 billion and aims to develop safe superintelligence.
“We have sufficient compute to prove, to convince ourselves and anyone else, that what we are doing is correct,” Sutskever stated confidently. SSI operates as “squarely an age of research company,” tackling AGI development with new approaches.
Emotions and Value Functions
The interview also discussed AI emotions and values. Sutskever cited the case of a patient with brain damage who lost the ability to feel emotions, reflecting on the role of emotions in human decision-making. He suggested that AI systems also need some form of “value function.”
Industry Impact
Sutskever’s statements could impact the entire AI industry:
Changes in Investment Strategy
- Potential shift from simple scaling investments to fundamental research investments
- Increased attention to startups with novel research ideas
Shifting Competitive Axes
- Research quality, not just compute resources, will determine competitiveness
- Architectural innovations like Google demonstrated with Gemini 3 become crucial
The Path to AGI
- AGI is not a simple extension of current approaches
- Fundamental research breakthroughs are required
Summary
Ilya Sutskever’s declaration of the “age of research” signals a turning point for the AI industry:
- Scaling limits: Building bigger models alone is insufficient
- Generalization challenge: Achieving human-like efficient learning is the next goal
- Research-first shift: New ideas and fundamental research become differentiators
The next phase of AI development will be a competition of wisdom, not compute.