Will Singularity Truly be Achieved in a Year’s Time? Bruce Rolff/Stocktrek Images – Getty Images
There is a flood of predictions regarding the timing of the singularity and the advent of artificial general intelligence (AGI). While some experts doubt its inevitability, others are eyeing 2026 as the milestone.
A recent comprehensive analysis of surveys conducted over the past 15 years sheds light on where scientists and industry professionals stand on the matter and how their perspectives have evolved, particularly in the wake of the emergence of extensive language models like ChatGPT. Despite a wide range of predictions spanning nearly half a century, the consensus is that AGI will materialize before the 21st century concludes.
Following the proliferation of large language models (LLMs) into various aspects of our digital existence, stakeholders from the scientific community, industry, and beyond have voiced their opinions on AI’s trajectory. Some researchers specializing in machine intelligence speculate that the singularity—when machines may surpass human intellect—might be only decades away. In contrast, the CEO of Anthropic believes we are on the brink of this breakthrough, with a timeline of approximately 12 months.
An analysis examining the forecasts of “8,590 scientists, prominent entrepreneurs, and the wider community” attempts to untangle the complexities of current AI prognostications and traces the evolution of these forecasts. This macro-level study was carried out by AIMultiple, a research organization known for its rigorous data analysis in evaluating new technologies. While various AI thresholds, including artificial general intelligence (AGI) and AI superintelligence, were considered, industry leaders generally displayed optimism in their predictions. Most respondents anticipate the advent of AGI within the next five decades.
However, with the introduction of LLMs in recent years, the timeline for AGI and the singularity has significantly shifted. “Current surveys of AI researchers predict AGI to emerge around 2040,” the report reveals. “Just a few years ago, before the rapid progress in large language models (LLMs), projections placed AGI around 2060. Entrepreneurs are even more bullish, foreseeing it around ~2030.”
The macro-analysis also provides insights into the inevitability of AGI as perceived by many experts. One key argument is that, unlike human intelligence, machine intelligence currently shows no discernible limits. With computing power doubling every 18 months (as per Moore’s Law), LLMs are expected to approach human-level computational capabilities swiftly. The report also suggests that in case traditional computing encounters obstacles, quantum computing could offer solutions.
“Most experts anticipate the end of Moore’s law in this decade,” the report notes. “Quantum computing’s distinctive attributes could efficiently train neural networks, the prevalent AI architecture in commercial applications. AI algorithms powered by stable quantum computers could potentially unlock the singularity.”
Nonetheless, not all experts are convinced of AGI’s inevit
The concept of intelligence is more intricate and diverse than what the current definition of Artificial General Intelligence (AGI) encompasses. Some AI experts view the human mind through the lens of eight intelligences, of which “logical-mathematical” is just one aspect (while others include interpersonal, intrapersonal, and existential intelligence). Yann LeCun, a pioneer in deep learning, advocates for a rebranding of AGI to “advanced machine intelligence,” arguing that human intelligence is too specialized to be replicated. Additionally, he suggests that while AI can be a valuable tool in facilitating new discoveries, it is unable to independently generate such discoveries. The report highlights that enhanced intelligence can result in more effectively designed and managed experiments, thereby increasing the potential for discoveries per experiment. It notes that even the most advanced machine, when analyzing existing data, may not possess the capability to discover a cure for cancer.
While individual predictions regarding the development of AGI span approximately half a century among experts and scientists, the overarching message remains clear: human society will inevitably undergo significant transformations due to these algorithms. The ultimate impact of these changes – whether positive or negative – is contingent upon our actions and decisions.
