Key takeaways
- 79% of researchers note a huge gap between public perception and reality.
- 74% believe current AI is driven by 'hype', slowing down real research.
- 76% think simple model scaling will not achieve AGI.
- Limited performance: Best current systems correctly answer only 50% of benchmark test questions.
The Major Gap with Science
The 2025 AAAI Presidential Panel Report (475 researchers, 24 studies) offers a pragmatic view far from media noise.
- 79% of researchers note a huge gap between public perception and reality.
- 74% believe current AI is driven by 'hype', slowing down real research.
- 76% think simple model scaling will not achieve AGI.
- Limited performance: Best current systems correctly answer only 50% of benchmark test questions.
Beyond the Hype Cycle
According to Rodney Brooks (MIT), we are in the heart of the disillusionment phase. It is urgent to change the paradigm to truly advance.
Priorities for Real Innovation
- Training methods favoring quality over quantity.
- Multi-agent systems checking each other for robustness.
- Standardization of ethics and governance.
Perception Paradoxes
While the public overestimates the proximity of AGI, they paradoxically underestimate current concrete AI gains in medicine, transport, and scientific research.
AH
Author
AI HUB Editorial
Research Desk
