The road to artificial general intelligence
SMRTR summary
Artificial intelligence that can discover drugs and write code still stumbles over puzzles that an average person solves in minutes. This cognitive disconnect reveals the central challenge in creating artificial general intelligence, or AGI.
Tech leaders are making bold predictions about timeline compression. Anthropic co-founder Dario Amodei suggests "powerful AI" with Nobel Prize-level thinking could emerge as soon as 2026, while OpenAI's Sam Altman sees AGI-like capabilities already "coming into view."
"Large language and reasoning models are transforming nearly every industry," notes Ian Bratt, vice president at Arm.
The forecast horizon keeps shrinking dramatically. Expert surveys now give at least a 50% chance of AI systems achieving several AGI milestones by 2028, with the timeline for machines outperforming humans in every possible task estimated at 50% by 2047.
What once seemed fifty years distant at GPT-3's launch has contracted to just five years by the close of 2024, reflecting both technological acceleration and heightened industry optimism about the convergence of training improvements, data availability, and computing power.
SMRTR provides this summary for quick context. The original article belongs to MIT Technology Review.
Read the original article