Subscribe to Our Newsletter

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks
SoftBank's Son Predicts Arrival of Superintelligent AI Within 10 Years
Source: SoftBank

SoftBank's Son Predicts Arrival of Superintelligent AI Within 10 Years

Philip Lee profile image
by Philip Lee

Tokyo, Japan—SoftBank Group Corp (9984.T) founder and CEO Masayoshi Son on Thursday unveiled an ambitious roadmap for artificial intelligence evolution. The roadmap predicts the emergence of superintelligent AI within ten years and outlines eight distinct phases of AI development.

Speaking at SoftBank World 2024, Son detailed his vision for AI progression:

  • Information processing
  • Knowledge understanding
  • Thinking and reasoning
  • Invention and creativity
  • Emotional understanding
  • Long-term memory development
  • Development of self-will
  • Superintelligence in harmony with humans

Son expects artificial general intelligence (AGI) to surpass human capabilities across multiple domains within two to three years, with artificial superintelligence (ASI) following within a decade.

"I have enough courage to say, within 10 years, 10,000 times of intellectual ability, which is super intelligence, should come," Son told attendees.

The billionaire entrepreneur highlighted recent advancements in AI, including OpenAI's latest model, which he claimed outperformed human experts in physics, chemistry, and biology exams at the Ph.D. level.

"GPT-4.0 answered correctly 56%, and human experts answered correctly more, but o1 got 78% of the correct score," Son said, referring to performance on complex scientific questions.

Son emphasized the distinction between artificial general intelligence (AGI) and artificial superintelligence (ASI). 

He defined AGI as AI achieving human-level performance across various tasks, while ASI would vastly exceed human capabilities.

The SoftBank chief predicted the adoption of personal AI agents within two to three years. 

These agents, he explained, would manage various aspects of daily life, from healthcare to financial planning.

"Personal agent is something that works closely with you. In two, three years, I think personal agent will become very common," Son said.

He also introduced the concept of "agent-to-agent" (AtoA) communication, envisioning a future where AI agents negotiate and make decisions for their human users.

Son stressed the importance of ethical considerations in AI development, stating that superintelligent systems should be designed to maximize human happiness and well-being.

"We need to wish for people's happiness. The biggest reward of the Q function should be designed like that, and then humankind will never be destroyed," he said, referring to reinforcement learning techniques used in AI training.

The SoftBank founder highlighted the potential of the "Internet of Things" (IoT) evolving into an "Agent of Things" (AoT), where AI agents would be embedded in everyday devices. 

He noted that SoftBank's chip design subsidiary, Arm Holdings, ships 30 billion chips annually, positioning the company to play a significant role in this transition.

Son's optimistic outlook comes as SoftBank invests heavily in AI and robotics companies through its Vision Fund, despite recent setbacks in some of its high-profile investments.

Industry analysts remain divided on the timeline for achieving human-level AI, with some experts warning of potential risks associated with rapid AI advancement. 

Given the complex challenges in developing truly general AI systems, critics argue that Son's predictions may be overly optimistic.

The speech also touched on the concept of "chain of thought" in AI development, emphasizing the importance of multi-step reasoning and ethical decision-making in advanced AI systems.

As global competition in AI development intensifies, Son's vision for the future of AI and SoftBank's investments in the sector will likely be closely watched by industry observers, policymakers, and competitors alike.

Philip Lee profile image
by Philip Lee

Subscribe to The Pickool

Success! Now Check Your Email

To complete Subscribe, click the confirmation link in your inbox. If it doesn’t arrive within 3 minutes, check your spam folder.

Ok, Thanks

Read More