The Road to Artificial General Intelligence: Forecasts and Milestones

Title: “The Pursuit of Artificial General Intelligence: Forecasts and Prospects”

For over a decade, Shane Legg, co-founder of the DeepMind AI lab now under Google, declared that the likelihood of achieving human-level artificial intelligence by 2028 stood at 50%.

This assertion from Legg appears steadfast as of his October 2023 interview with tech program host Dwarkesh Patel, where he maintained researchers still had a 50% chance of attaining general artificial intelligence by 2028.

Similarly, computer scientist and futurist Raymond Kurzweil anticipates AI surpassing the Turing Test by 2029, reaching a level comparable to human general intelligence.

DeepMind, under Google, introduced a novel categorization of AI levels in November 2023:

1) Nascent AI, like chatbots.
2) Competent AI.
3) Expert AI.
4) Gifted AI.
5) Superintelligent AI, capable of a wide array of tasks, surpassing human capabilities, including foreseeing future events, communicating with animals, and deciphering human thought patterns.

However, DeepMind’s team indicates that AI has yet to surpass the nascent level of general artificial intelligence.

Ibn Khaldun, over 650 years ago, posited that “the future can be predicted based on scientific principles.” Will we achieve superintelligent AI, and can it fulfill the aforementioned predictions by 2030 or sooner?