Get your free personalized podcast brief
We scan new podcasts and send you the top 5 insights daily.
Huang redefines "superintelligence" not as a single, all-knowing AGI, but as specialized systems that vastly outperform humans at a specific task. Citing self-driving cars and protein synthesis as examples, he asserts that we have already crossed the threshold into the era of superintelligent AI in narrow, practical applications.
Related Insights
Jensen Huang criticizes the focus on a monolithic "God AI," calling it an unhelpful sci-fi narrative. He argues this distracts from the immediate and practical need to build diverse, specialized AIs for specific domains like biology, finance, and physics, which have unique problems to solve.
Framing AGI as reaching human-level intelligence is a limiting concept. Unconstrained by biology, AI will rapidly surpass the best human experts in every field. The focus should be on harnessing this superhuman capability, not just achieving parity.
OpenAI's CEO believes the term "AGI" is ill-defined and its milestone may have passed without fanfare. He proposes focusing on "superintelligence" instead, defining it as an AI that can outperform the best human at complex roles like CEO or president, creating a clearer, more impactful threshold.
"Superintelligence" is clearly defined as AI that is better, faster, and cheaper than the best humans at everything. In contrast, "AGI" (Artificial General Intelligence) is a vague term for general-purpose AI, a milestone that current models have arguably already achieved.
AI's capabilities are highly uneven. Models are already superhuman in specific domains like speaking 150 languages or possessing encyclopedic knowledge. However, they still fail at tasks typical humans find easy, such as continual learning or nuanced visual reasoning like understanding perspective in a photo.
Benchmarks like GDPVal show models like GPT-4 consistently outperform human experts on professional tasks, meeting the practical definition of AGI for knowledge work. The public discourse, however, has prematurely shifted the goalposts to sci-fi concepts of Artificial Superintelligence (ASI), obscuring the revolution already underway.
The next leap in AI will come from integrating general-purpose reasoning models with specialized models for domains like biology or robotics. This fusion, creating a "single unified intelligence" across modalities, is the base case for achieving superintelligence.
The path to AGI won't be uniform. Instead, we'll see 'jagged superintelligence,' where models achieve superhuman capabilities in specific verticals with high verifiability, such as coding, finance, and scientific research. These specialized peaks of excellence will appear long before a generalized intelligence is achieved.
A paper co-authored by DeepMind's Chief AGI Scientist offers a new benchmark for superintelligence (ASI): a system that outperforms large organizations of thousands of experts working over extended periods, reframing the goalpost beyond individual human genius.
Yann LeCun posits that the goal of AI should not be to replicate the breadth of human intelligence (AGI). Instead, development should focus on creating specialized models that achieve superhuman depth in fields like physics and chemistry, as this is where true breakthroughs will occur.
