Artificial General Intelligence (AGI) has transitioned from an abstract concept to a focal point in the tech industry recently, sparking intense debate among experts. The rise of AGI is a matter of when, not if. UPP’s recent updates suggest that AGI could possibly emerge as early as 2026, though opinions vary widely.
What is Artificial General Intelligence (AGI)?
Artificial general intelligence (AGI) is an area of artificial intelligence (AI) research in which scientists aim to create a computer system that is generally smart as or smarter than humans. These AGI systems could be self-understanding and self-control – including the ability to edit and process their own code - and be able to learn to solve problems like humans autonomously.
The term "Artificial General Intelligence" was first introduced in a 2007 essay collection edited by computer scientist Ben Goertzel and AI researcher Cassio Pennachin. However, the idea has been around for decades, shaping AI's historical development and frequently appearing in science fiction literature and films.
However, in 2025, signals about the arrival of AGI and the rising topics about it are likely to appear. “I believe that the right time to start preparing for A.G.I. is now”, said by Kevin Roose - a technology columnist and a co-host of the New York Times tech podcast “Hard Fork”.
Predictions About AGI From Experts Through Recent Surveys
Reputable surveys in the technology industry are also showing many “conflicting” opinions and predictions from AI expert, generally in 3 different ways:
In the first way, a survey of 475 AI researchers revealed that approximately 76% believed it was "unlikely" or "very unlikely" that simply expanding current AI methods would lead to AGI. These results were included in a report by the Association for the Advancement of Artificial Intelligence, a global scientific organization headquartered in Washington, DC.
Surprisingly in the second way, a recent analysis conducted on February 18 by Cem Dilmegani, principal analyst at AIMultiple Research, examined around 8,600 expert predictions from scientists, AI specialists, and entrepreneurs made between 2009 and 2023 to determine when AGI might become a reality. As part of this study, a subset of data from 10 surveys, involving responses from 5,288 AI researchers and experts, was analyzed. By averaging the findings, the study estimated a 50% possibility that machines could achieve human-level intelligence sometime between 2040 and 2061.
However, in the third way, recent surveys predict that the technological singularity could occur sooner than previously expected. A 2023 study, which surveyed 2,778 scientists, indicated that AGI would likely be achieved by 2040 at the latest. However, some experts, such as Dario Amodei, AI researcher and CEO of Anthropic, suggest it could happen as early as 2026. Goertzel also suggested we may reach the singularity by 2027, while the co-founder of DeepMind, Shane Legg, has said he expects AGI by 2028. Musk has also suggested AI will be smarter than the smartest human by the end of 2025.
AGI Benchmark Created To Test AI Models
To deal with the rising concerns of AGI, scientists are continuously developing new 'AGI benchmark' to test the capability of new AI models and likely to predict the chance for AGI to appear soon or later.
The Arc Prize Foundation, a nonprofit co-founded by renowned AI researcher François Chollet, announced in a blog post on Monday the launch of a new, challenging test designed to evaluate the general intelligence of advanced AI models. This new test, called ARC-AGI-2, has proven to be a significant challenge for most AI systems. According to the Arc Prize leaderboard, reasoning-based AI models such as OpenAI’s o1-pro and DeepSeek’s R1 achieved scores ranging from 1% to 1.3% on ARC-AGI-2. Meanwhile, powerful non-reasoning models, including GPT-4.5, Claude 3.7 Sonnet, and Gemini 2.0 Flash, scored around 1%. To establish a human baseline, the Arc Prize Foundation had over 400 individuals take the ARC-AGI-2 test. On average, these participants correctly answered 60% of the questions - significantly outperforming any AI model tested so far.
Before the ARC-AGI-2 of The Arc Prize Foundation, researchers at OpenAI have developed MLE-bench, a set of 75 highly challenging tests designed to evaluate whether a future advanced AI system can modify its own code and enhance its capabilities. The benchmark, known as "MLE-bench," consists of 75 Kaggle challenges, each designed to assess machine learning engineering skills. These tasks involve training AI models, preparing datasets, and conducting scientific experiments, with the Kaggle tests evaluating how effectively machine learning algorithms perform on specific problems.
So far, big tech companies and AI experts are putting much effort into developing benchmarks to test out the potential of recent AI models.
Conclusion about AGI
While many tech companies present AGI as their ultimate objective, its exact definition remains unclear. Google DeepMind defines AGI as a system capable of surpassing human performance on a range of cognitive tests, whereas Huawei argues that true AGI requires a physical body to interact with its environment. Meanwhile, an internal report from Microsoft and OpenAI suggests they will recognize AGI as achieved only when OpenAI develops a model capable of generating $100 billion in profit.
Productized AI With UPP Global Technology JSC
As the race toward Artificial General Intelligence (AGI) accelerates, businesses must find practical ways to integrate cutting-edge AI advancements into real-world applications. UPP Global Technology JSC offers “Productize AI”, a service designed to help enterprises transform AI research and emerging technologies into scalable, market-ready solutions. Whether leveraging large language models (LLMs), automating complex workflows, or developing AI-driven decision-making systems, UPP Global Technology JSC ensures that organizations can harness the power of AI effectively.
Related article:
Edge AI Explained: The Next Frontier of Technology (Part 2) | UPP
Edge AI Explained: The Next Frontier of Technology | UPP
References:
O'Neill, M. (2025, March 8). AGI could now arrive as early as 2026 — but not all scientists agree. Live Science. https://www.livescience.com/technology/artificial-intelligence/agi-could-now-arrive-as-early-as-2026-but-not-all-scientists-agree
New Scientist Staff. (2025, March 14). AI scientists are skeptical that modern models will lead to AGI. New Scientist. https://www.newscientist.com/article/2471759-ai-scientists-are-sceptical-that-modern-models-will-lead-to-agi
O'Neill, M. (2024, September 20). What is artificial general intelligence (AGI)? Live Science. https://www.livescience.com/technology/artificial-intelligence/what-is-artificial-general-intelligence-agi
Loten, A. (2025, March 24). A new challenging AGI test stumps most AI models. TechCrunch. https://techcrunch.com/2025/03/24/a-new-challenging-agi-test-stumps-most-ai-models/
O'Neill, M. (2024, October 14). Scientists design new AGI benchmark that may say whether any future AI model could cause catastrophic harm. Live Science. https://www.livescience.com/technology/artificial-intelligence/scientists-design-new-agi-benchmark-that-may-say-whether-any-future-ai-model-could-cause-catastrophic-harm