While current AI excels at specialized tasks such as drug discovery and code generation, it often falters with problems easily solved by humans. This discrepancy underscores the challenge of achieving Artificial General Intelligence (AGI): developing AI with human-level or greater intelligence across all domains.
Predictions for AGI arrival vary. Anthropic co-founder Dario Amodei foresees “powerful AI” as early as 2026, potentially boasting Nobel Prize-caliber intellect, versatile interface capabilities, and independent reasoning. OpenAI’s CEO Sam Altman suggests AGI-like features are already emerging, fueled by advancements in training methodologies, data availability, computational power, and decreasing costs.
Expert consensus indicates a significant probability of achieving key AGI milestones by 2028. One survey estimates a 10% chance of completely unaided machines surpassing human capabilities in all tasks by 2027, with a 50% likelihood by 2047. This timeline is continuously recalibrated with each new breakthrough in the AI field.
Ian Bratt, VP at Arm, highlights the transformative impact of large language and reasoning models across diverse industries.