2024 Predictions Regarding Super-Intelligent AI Now Feel Real Ahead of 2027
The development of artificial intelligence (AI) in mid-2ability 2026 is moving at a terrifying and almost nonsensical speed. Two years ago, while the public was still mesmerised by the initial capabilities of Large Language Models (LLMs), a stern warning emerged from the heart of the technology industry itself.
In mid-2024, a former OpenAI researcher released a manifesto that exposed the reality behind leading AI laboratories. He predicted that artificial intelligence systems would achieve the level of Artificial General Intelligence (AGI) by 2027 at the latest. This phenomenal 165-page manifesto, titled ‘Situational Awareness: The Decade Ahead’, was written by Leopold Aschenbrenner.
Aschenbrenner is no mere observer; he is a former member of the Superalignment team at OpenAI, a specialised division formed by founder Ilya Sutskever. The division’s mission was to ensure that future AI would not defect, spiral out of control, or endanger human existence. Now, in 2026, the document once dismissed as an ‘empty prophecy’ has become essential reading for tech executives, investors, and policymakers in Washington.
As 2027 approaches, an in-depth analysis of the ‘Situational Awareness’ report reveals how these warnings are manifesting. The first part of the essay focuses on what Aschenbrenner calls ‘Counting the OOMs’ (Orders of Magnitude). He breaks down the fact that the AI industry is not moving by chance, but is progressing along a highly precise, mathematical, and predictable exponential growth path.
While many laypeople view the leaps in intelligence from GPT-2 in 2019, to GPT-3, and then to GPT-4 in 2023 as mere technological miracles, Aschenbrenner refutes this view. Such drastic increases in capability are purely the result of a simple mathematical equation: the massive addition of computing power combined with continuously refined algorithmic efficiency. He calculates that every year, AI models become significantly smarter because the training data and GPU processing power injected into them increase exponentially in terms of OOM (multiples of ten). If the trend from the last five years is projected forward, he convincingly claims that the threshold of human-expert-level intelligence (AGI) will be definitively breached by 2027. Now, with the OOM leaps being continuously driven by tech companies, the 2027 target for human-level intelligence is no longer mere discourse, but a deadline that is only months away. One of the greatest misconceptions circulating in society is the assumption that once AGI is finally achieved, technological innovation will hit a ‘limiting wall’, reach a point of saturation, and slow down.
Source: jawawa.id (https://jawawa.id/newsitem/2024-predictions-regarding-super-intelligent-ai-now-feel-real-ahead-of-2027-1780882174)
Originally published by Kompas (https://tekno.kompas.com/read/2026/06/01/09090097/prediksi-tahun-2024-soal-ai-super-cerdas-kini-terasa-nyata-jelang-2027?utm_source=ABC&utm_medium=rss), translated from Indonesian