Super intelligence can be born in 2026

The appearance of artificial super intelligence, also known as general artificial intelligence (AGI), the AI ​​system with superior intelligence can perform well the diverse tasks, just a matter of time, according to the results of analysis of thousands of experts, Live Science On 9/3 reported. The new analysis updated by Cem Dilmegani, analyst at Aimultiple Research, conducted a review of about 8,600 predictions from scientists, AI experts and business experts from 2009 to 2023 to find out when AGI was born.

A part of analysis includes 10 surveys with a total of 5,288 researchers and AI experts. Based on the average calculation from data, 50% of the possibility of machinery will reach human intelligence between 2040 – 2061. The survey recently more than predicting the strange technology scores can come sooner. For example, one of the latest studies conducted in 2023 interviewed 2,778 scientists and pointed out we could achieve AGI earliest in 2040. Some experts in this field such as Dario Amodei, AI researcher and CEO

The arising of AGI is motivated by the rapid progress of a large language -based language model (LLM). Transformer is a Neural network architecture designed to process and understand natural language data (NLP) effectively. This is the technology used on chatboxes like chatgpt and image creation like Dall-E. Before these technologies, in 2019, some scientists predicted that AGI could occur in 2060 or never.

Analysis provides several reasons why scientists think AGI will certainly happen. First, unlike human intelligence, there is no theoretical limit to computer capacity. It is according to the law of Moore predicting the power will double every 18 months. The future AI systems will one day compare to the human intelligence in terms of calculating each second. In recent years, many experts believe that the law of Moore is no longer true.

The study also refers to quantum computers as a means to overcome the computer limit. Quantum computers can handle multiple calculations simultaneously by exploiting quantum mechanical theorems. Classic computers, including the fastest supercomputers, must be calculated in order. But quantum computers can lead to high -end AI systems with faster processing capacity than the best models today.

However, some other scientists in the field think that it is necessary to have a further breakthrough before we can get closer to AGI. For example, Yann Lecun, Facebook’s AI science director, identifies a transformer -based structure and the current approach to AI is not compatible with human intelligence.