Sam Altman recently stated that OpenAI has found the direction toward achieving AGI, and they may achieve AGI by 2025
OpenAI CEO Sam Altman had an indepth conversation with YC President Garry Tan. Sam Altman shared his views on AGI development. The most surprising was Sam Altman's statement: AGI will arrive in 2025, and ASI will emerge within thousands of days.
What exactly are the AGI and ASI that Altman talked about?
Artificial intelligence development has different stages, and based on intelligence levels, it can be divided into three stages: Artificial Narrow Intelligence (ANI), Artificial General Intelligence (AGI), and Artificial Super Intelligence (ASI).
Let's understand each stage in detail:
- Artificial Narrow Intelligence (ANI)
Characteristics:
Can only perform specific tasks
Excels in specific domains
No real understanding or learning ability
Relies on preset programs and data
Cannot handle problems outside its set parameters
- Artificial General Intelligence (AGI)
Characteristics:
Has humanlike intelligence
Can understand and learn any intellectual task
Possesses reasoning and planning abilities
Can perform abstract thinking
- Artificial Super Intelligence (ASI)
Characteristics:
Surpasses human intelligence
Superior to humans in all aspects
Has selfimprovement capabilities
In simple terms, these stages are like the growth process of artificial intelligence: from being able to do just one thing, to gradually being able to do many things, and finally possibly developing its own thoughts. Current AI is still in its early stages, like a child, continuously growing.
AI may enter the AGI growth stage in 2025
Currently, AI is still in a relatively early stage. However, in his latest interview, Altman discussed "AGI (Artificial General Intelligence)." He predicted that AGI will become a reality in 2025, and ASI will arrive within thousands of days.
Why is Altman so confident? Let's look at what he said in the interview:
Sam Altman stated: "My company now knows how to achieve artificial general intelligence. This is the first time I felt we really know how to do it. For example, I think building AGI from now still requires a lot of work. While there are some known unknowns, I think we basically know how to do it. It will take time and be difficult, but it's very exciting."
"I think there are more problems to solve on the product side, but roughly speaking, we know what to aim for and what we want to optimize. This is a very exciting moment. When you have this kind of clarity, I think you can progress very quickly. If you're willing to say, we're going to do these few things, we'll work hard to do them well. Our research path is quite clear. Our infrastructure path is quite clear. Our product path is becoming clearer. You can position yourself well around this."
How credible is the likelihood of achieving AGI by 2025?
Many people question whether Sam Altman is just hyping things up.
However, Noam Brown, a senior researcher at OpenAI known as the "father of poker AI," publicly stated: "Based on my firsthand experience and observations at OpenAI, Altman's views are not exaggerated and actually align very well with the general views of the company's frontline researchers."
Elon Musk previously predicted AGI would arrive by 2029. In his most recent interview, he predicted that AI might surpass the smartest humans before next year or 2026. Within the next five years, AI capabilities might exceed all of humanity.
Such confirmations make Altman's statements seem more worthy of attention.
Some Speculation
Imagine if artificial intelligence really becomes as smart as humans and can do all sorts of things it would bring revolutionary changes.
AGI can think, learn, create like humans, and complete all human work. This change also raises concerns: will many people lose their jobs as a result? Will AI decisions always meet ethical standards? So how to enjoy the convenience brought by AI while avoiding potential problems is also worth considering.
There will always be people who think predictions about AGI's arrival are just hype and false. But undeniably, AI technology is evolving rapidly with quick updates and iterations. TakeXXAI as an example to improve user experience, it constantly upgrades and updates, reaching the next stage in just one week.
Not to mention that major AI giants have never stopped researching artificial intelligence, so we may really welcome AGI in 2025.