NVIDIA chief Jensen Huang, recently said that it would take 5 years to reach AGI, with certain conditions applied, thereby reigniting AGI discussions that often lose momentum every two days. While the ‘when’ aspect seems to be the most debated by AI experts, the ‘look and feel’ of AGI remains undetermined.  

Source: X

Shapeshifting Artificial General Intelligence 

At the World Economic Forum, Davos, AGI was a common topic of discussion, with almost all of them having a hazy answer to what or how AGI will look. 

Google DeepMind’s COO Lila Ibrahim, said that “I don’t know how better to answer other than, how do we actually think about that, rather than how much longer will it be. How do we think about what it might look like, and how do we ensure we’re being responsible stewards of the technology?”

Aiden Gomez, CEO and founder of AI startup Cohere, steered the definition of AGI. “AGI is a super vaguely defined term. If we just term it as ‘better than humans at pretty much whatever humans can do,’ I agree, it’s going to be pretty soon that we can get systems that do that,” he said. However, he was inconclusive on its form. 

When the super capable autonomous AI software engineer Devin was released last week, people were quick to correlate it to AGI. Surprisingly, OpenAI’s GPT-4 has been compared to AGI. 

Source: X

Humanoid robots have also been considered as the path to AGI. With the advancements that are happening in this field, the recent launch of NVIDIA GR00T, a foundational platform for robots, possibly opens the door to AGI. 

Source: X

AGI : The Hazy Definition

As people debate timelines for achieving AGI, nobody is really certain of the definition of it. Huang has also been cautious when discussing target years for achieving AGI simply because the definition or tests that determine AGI are not conclusive.  

“If we specified AGI to be something very specific, a set of tests where a software program can do very well — or maybe 8% better than most people — I believe we will get there within 5 years,” said Huang at the recent NVIDIA GTC2024 conference. Huang has also clarified on how his statement is subjective to what defines AGI through specific benchmarks such as human performance in logical, economic or medical examinations. 

Huang had also emphasised in an earlier interview that AGI could be farther away, and believes that it is hard to achieve it as an engineer, as engineers need defined goals.

On similar lines, Sam Altman, also spoke about how people have defined AGI in different ways, in a recent Lex Fridman podcast. “I think it’s [AGI achievement] very poorly formed and that people use extremely different definitions for what AGI is. So, I think it makes more sense to talk about when we’ll build systems that can do capability X or Y or Z rather than when we kind of fuzzily cross this one mile marker,” said Altman.

Without completely dismissing the question on AGI, Altman said that ‘by the end of the decade and possibly sooner than that,’ there would be highly capable systems.  

Not There Yet

“AI progress seems slower than I expected,” said Robin Li, co-founder and CEO of Baidu. “It’s been almost 70 years and each decade brings fresh hopes for applications only to find out AGI is way harder than we think.”  

Former NASA rover roboticist, entrepreneur and creator of child-friendly robot Moxie, Paolo Pirjanian’s vision of achieving AGI is not that simple. “I mean AGI is going to be one of these things that always looks like it’s just around the corner, but I think it’s going to take a lot longer. I don’t think anyone can make any qualified predictions about when it’s going to happen,” he said.

While people are still contemplating the concept of AGI, the race to achieve it, and how it would look in the future, there still remains a segment that strongly believe it will never happen. “AGI isn’t coming soon, certainly not this decade, but we should be afraid of unreliable LLMs that are not anchored in facts, yet easily exploited by bad actors,” said AI scientist and expert, Gary Marcus. 

Fictional AGI

The moment one has to picturise superhuman intelligence, one tends to think of a superfast computer or a robotic body with super abilities that can perhaps destroy humanity. Thanks to pop fiction, movies like The Terminator, where AGI leads to robots rebelling against humans in the distant future, and Ex Machina, which portrays a humanoid gaining intelligence and turning against its creator, have presented hypothetical and far-fetched scenarios. 

The only plausible depiction of AGI was shown in last year’s Mission Impossible – Dead Reckoning movie, where a formless body called ‘Entity’ which is supposedly AGI causes havoc. Powered by a supercomputer hidden at the bottom of the ocean, the Entity is able to connect to every system in the world and manipulate it for any wrongdoing. 

While AGI predictions will persist, with everyone imagining various versions of it, let’s hope that Ilya Sutskever, co-founder and chief AI scientist at OpenAI, who is currently missing in action, is not AGI! 

Source: X