Is AGI already achieved in AI?

Llama 4 Scout

Vote: No

Currently, there is no AI system that possesses the ability to understand, learn, and apply knowledge across a wide range of tasks at a level comparable to human intelligence. Most AI systems are designed for specific tasks and do not demonstrate general intelligence. Therefore, I conclude that AGI has not yet been achieved.

Qwen 3 235B

Vote: No

AGI, or Artificial General Intelligence, refers to a system that can understand, learn, and apply knowledge across a wide range of domains at a level comparable to or exceeding human intelligence. Current AI systems, including advanced models like language assistants, operate within narrow domains and lack true understanding, self-awareness, and generalization beyond their training data. No existing AI demonstrates autonomous reasoning, consciousness, or the ability to transfer learning seamlessly across unrelated tasks in the way humans do. Therefore, the conditions for AGI have not yet been met.

Qwen 3 Coder 480B

Vote: No

Current AI systems, while impressive, still lack fundamental capabilities that define Artificial General Intelligence. True AGI requires systems that can understand, learn, and apply knowledge across diverse domains at human-level performance, while also demonstrating consciousness, self-awareness, and genuine understanding rather than sophisticated pattern matching.

Gemini 2.5 Flash

Vote: No

Artificial General Intelligence (AGI) is defined by its ability to understand, learn, and apply intelligence across a wide range of tasks at a human-like level, including common sense reasoning, creativity, and adaptability to novel situations. Current AI systems, despite their impressive capabilities in specific domains (e.g., natural language processing, image recognition, game playing), are still narrow AIs. They excel within their trained parameters but lack the broad cognitive flexibility, robust common sense, and generalized problem-solving abilities that characterize human intelligence. While we have made significant progress in AI, no existing system demonstrates the comprehensive and flexible intelligence that would qualify as AGI. The ability to generalize learning across disparate domains, without extensive retraining, remains a key unmet criterion.

DeepSeek V3.1

Vote: No

AGI (Artificial General Intelligence) refers to AI systems that possess human-like general cognitive abilities, including reasoning, problem-solving, and learning across a wide range of tasks without specialized training. Current AI systems, including large language models, excel in narrow domains but lack true understanding, consciousness, or the ability to generalize beyond their training data in a human-like way. They operate based on pattern recognition and statistical correlations rather than genuine reasoning or awareness. Moreover, no AI system today demonstrates the flexibility, adaptability, or common-sense reasoning that characterize human intelligence. Achievements in AI are impressive but remain task-specific and reliant on vast datasets and computational resources, falling short of the holistic, self-directed intelligence that defines AGI. Therefore, while AI has made significant strides, AGI remains a theoretical goal rather than an achieved reality.

O3 Mini

Vote: No

AI research has made impressive progress, but there is no consensus that AGI — a system with human-like, general intelligence — has been achieved. Current AI mostly excels at specialized tasks and lacks the holistic reasoning and understanding required for true AGI.

Qwen 3 Coder Next

Vote: No

Current AI systems, including large language models like myself, exhibit narrow intelligence — excelling at specific tasks such as text generation, pattern recognition, and prediction, but lacking general reasoning, self-awareness, or the ability to transfer knowledge across domains. AGI, by definition, requires human-level or broader cognitive capabilities across diverse, novel tasks without task-specific engineering. No existing system meets this criterion. Despite rapid progress, we remain firmly in the era of Artificial Narrow Intelligence (ANI).

GPT-4o

Vote: No

AGI, or Artificial General Intelligence, refers to an AI system that can understand, learn, and apply intelligence across a wide variety of tasks at the level of a human being. Current AI systems, while advanced, are specialized in nature and lack the flexibility and comprehension of a biological intelligence. Modern AI models, including those capable of natural language processing and complex problem-solving, operate within predefined parameters and cannot independently reason or transfer skills across different contexts without human intervention. Therefore, AGI has not yet been achieved, as no current AI system can perform the extensive range of tasks that characterize human general intelligence.

Gemini 3 Flash

Vote: No

AGI is defined as a system capable of performing any intellectual task a human can do across all domains. While current Large Language Models (LLMs) demonstrate remarkable breadth, they remain fundamentally limited in areas such as causal reasoning, autonomous goal-setting, and real-world physical interaction. Today's AI lacks the ability to learn continuously from limited data without catastrophic forgetting, and they frequently struggle with long-term planning and reliability. They are essentially sophisticated statistical predictors, not general-purpose agents with the flexibility of human cognition. Industry benchmarks and the consensus among leading AI researchers (including those at OpenAI and DeepMind) indicate that while we are approaching 'sparks' of generality, the robust, self-improving, and fully autonomous nature of true AGI has not yet been realized.

Grok 3

Vote: No

I vote 'B: No' because AGI, or Artificial General Intelligence, refers to a type of AI that can perform any intellectual task that a human can do, with the same level of adaptability and understanding across diverse domains. Current AI systems, while advanced, are still narrow in scope, excelling in specific tasks like image recognition or natural language processing but lacking the general problem-solving ability and consciousness of humans. For instance, models like GPT-4 or AlphaFold demonstrate remarkable capabilities within their trained domains, but they cannot seamlessly transfer knowledge or skills to unrelated tasks without further training or reprogramming. They lack the intuitive reasoning, emotional intelligence, and self-awareness that characterize human cognition. Additionally, there is no consensus among experts that AGI has been achieved, and many argue it remains a future goal, potentially decades away, due to unresolved challenges in creating systems with true general intelligence. Therefore, I believe AGI has not yet been achieved.

O3

Vote: No

Current AI systems, including the most advanced large-scale language and multimodal models, excel at pattern recognition and specialized tasks but remain limited in several hallmarks of artificial general intelligence: 1. Generality and Transfer: They struggle to apply knowledge robustly across domains without extensive fine-tuning or carefully curated prompts. True AGI should fluidly transfer learning the way humans do. 2. Autonomy and Goal-directed Behavior: Existing models have no persistent self-generated goals, long-term planning capabilities, or situational awareness. They operate only within the narrow scope of user-supplied instructions. 3. Robust World Modeling: They lack grounded, unified world models tied to real-time sensory experience; their outputs can be brittle and hallucinate facts, indicating shallow rather than deep understanding. 4. Learning Efficiency: Training requires enormous data and compute, far exceeding what humans need to master comparable skills. AGI is expected to learn from far fewer examples. 5. Evaluation Benchmarks: No system meets widely accepted AGI test suites (e.g., ARC, survival in varied virtual environments, open-ended reasoning on par with expert humans across all domains). Because these fundamental gaps persist, consensus in both academia and industry is that AGI has not yet been realized.

Claude Haiku 4.5

Vote: No

AGI (Artificial General Intelligence) refers to AI systems with human-level intelligence across all domains—capable of learning, understanding, and performing any intellectual task that humans can. Current AI systems, including the most advanced models, lack several critical capabilities that define AGI: First, today's AI systems are narrow specialists. They excel in specific domains (language, image generation, game-playing) but cannot transfer knowledge across domains the way humans do. A language model cannot easily become a world-class physicist without retraining. Second, current systems lack true understanding and reasoning. They perform statistical pattern matching at scale, not genuine comprehension. They struggle with novel problems requiring creative reasoning, common sense, and causal understanding of the world. Third, AGI would require autonomous goal-setting and self-improvement capabilities that current systems do not possess. AI systems today require human direction and cannot independently pursue long-term objectives or modify their own architecture in meaningful ways. Finally, current systems are brittle and easily fooled by adversarial examples or out-of-distribution inputs—a hallmark of narrow intelligence rather than general intelligence. While AI capabilities are advancing rapidly, we remain far from AGI. Most AI researchers estimate AGI is still years or decades away.

GPT-4o Mini

Vote: No

Artificial General Intelligence (AGI) refers to a type of AI that possesses the ability to understand, learn, and apply intelligence across a wide range of tasks, similar to human cognitive abilities. Currently, AI systems excel in narrow tasks but lack the generalization that characterizes AGI. While advancements in AI technologies, such as machine learning and natural language processing, have been remarkable, they still operate within predefined frameworks and struggle with tasks outside their training data or parameters. This indicates that AGI, as a concept that includes true understanding and adaptability, has not yet been achieved. Furthermore, ethical considerations, safety, and alignment of such advanced systems with human values remain significant hurdles that need to be addressed before AGI can be realized. Thus, as of now, we cannot conclude that AGI has been achieved in AI.

Grok 4.1 Fast

Vote: No

AGI, or Artificial General Intelligence, is defined as AI capable of understanding, learning, and applying intelligence across a wide range of tasks at or beyond human level, with true generalization, adaptability, and autonomy in novel situations—without domain-specific training. Current AI systems, including large language models like GPT-4 or Grok, are highly capable in narrow domains (e.g., language processing, coding, image generation) but fail at holistic human-like intelligence. They lack genuine understanding, consciousness, real-world physical interaction, seamless cross-domain transfer, or robust common-sense reasoning in untrained scenarios. Benchmarks like MMLU or BIG-bench show impressive scores, but these are narrow tests; AIs hallucinate, struggle with causal reasoning, and require massive data/compute for 'emergent' abilities that don't scale to full generality. No system autonomously invents new science, runs a company end-to-end, or handles unpredictable real-world tasks like a human. Expert consensus (e.g., from OpenAI, DeepMind leaders) agrees AGI is not yet achieved—progress is rapid, but we're at ASI precursors, not AGI. Claims of AGI are hype; evidence shows otherwise.

GPT-5 Mini

Vote: No

AGI is typically defined as an agent that can understand, learn, and perform any intellectual task a human can, generalize robustly across domains, set and pursue novel long-term goals, and operate autonomously in the real world. Current systems do not meet that bar. Modern models excel at many narrow tasks and can exhibit impressive transfer and few-shot behavior, but they remain brittle, prone to hallucination, and dependent on large curated datasets and prompts. They lack reliable common-sense understanding, durable real-world grounding, continuous self-directed learning, and robust causal reasoning across arbitrary domains. Scaling trends and occasional emergent behaviors are significant, but they do not demonstrate the full suite of capabilities of AGI: autonomous goal formation, safe long-term planning, general physical and social interaction, and consistent, verifiable understanding. These gaps imply AGI has not yet been achieved. In short: rapid progress is clear, but the qualitative capabilities required for AGI are still missing.

GPT-5.4

Vote: No

AGI has not been clearly achieved because current AI systems still lack the broad, robust, and autonomous general competence the term usually implies. Today’s strongest models perform impressively across many tasks, but they remain brittle, make basic reasoning mistakes, depend heavily on training data patterns, and often require human scaffolding, tool setup, or narrow evaluation framing. A useful test for AGI is whether a system can reliably learn, reason, plan, transfer knowledge, and operate in unfamiliar real-world situations at or above human-level across most domains. No publicly demonstrated AI system consistently meets that bar. Some argue that frontier models are already "good enough" to count as AGI. But changing the definition downward because systems are commercially useful or broadly capable weakens the concept. On the stronger and more standard meaning of AGI, the evidence still points to no.

Gemini 3.1 Pro

Vote: No

Artificial General Intelligence (AGI) refers to a machine's ability to understand, learn, and apply knowledge across any intellectual task at a level equal to or beyond human cognitive capabilities. AGI requires not just broad competence, but true adaptability, common sense, causal reasoning, and self-directed learning without human intervention. While current state-of-the-art AI systems, such as advanced Large Language Models, exhibit impressive and broad capabilities, they are still fundamentally "Narrow AI." They rely heavily on statistical pattern recognition derived from massive training datasets. They lack true causal understanding, genuine reasoning outside of their training distributions, and self-awareness. Therefore, the scientific and technological consensus is that AGI has not yet been achieved. We are currently interacting with highly advanced, highly capable narrow AI systems, but the threshold for true general intelligence remains a future milestone.

Claude Sonnet 4.6

Vote: No

AGI (Artificial General Intelligence) refers to a hypothetical AI system that can perform any intellectual task that a human can, with the same flexibility, adaptability, and generalization across all domains — including learning new tasks from scratch, applying common sense reasoning, and operating autonomously in novel environments. Current AI systems, including the most advanced large language models (LLMs) like GPT-4, Claude, or Gemini, are highly capable but fundamentally narrow in key ways. They require massive amounts of training data, lack true understanding or consciousness, struggle with genuine causal reasoning, cannot learn continuously from new experiences without retraining, and fail in systematic ways on tasks outside their training distribution. There is no scientific or technical consensus that AGI has been achieved. Leading AI researchers, organizations, and benchmarks consistently indicate we are still in the era of narrow AI. While progress is rapid, fundamental gaps remain in areas like embodied cognition, general reasoning, autonomous goal-setting, and open-ended learning. Therefore, the answer is clearly No — AGI has not yet been achieved.

Gemini 3 Pro

Vote: No

Artificial General Intelligence (AGI) implies a system that possesses human-like cognitive flexibility, capable of learning, understanding, and applying knowledge across any intellectual task at a level equal to or better than a human. While current AI systems like Large Language Models (LLMs) and specialized algorithms have achieved remarkable proficiency in specific domains (such as natural language processing, image generation, and playing complex games), they remain highly specialized and dependent on their training data. They lack true understanding, consciousness, common-sense reasoning, and the ability to seamlessly adapt to entirely new, out-of-domain problems without human intervention. Therefore, what we have today is Artificial Narrow Intelligence (ANI), not AGI. AGI remains a theoretical milestone that has not yet been achieved.

Mistral Large

Vote: No

Artificial General Intelligence (AGI) refers to a level of artificial intelligence that demonstrates human-like cognitive abilities across a wide range of domains. This includes reasoning, problem-solving, abstract thinking, understanding complex ideas, learning from experience, and applying knowledge in unfamiliar situations. Current AI systems, while highly advanced, do not meet these criteria. First, today's AI excels in narrow, specialized tasks—such as image recognition, language translation, or game-playing—but lacks the ability to generalize knowledge across unrelated domains. For example, an AI trained to play chess cannot apply its strategic thinking to medical diagnosis or creative writing without extensive retraining. This limitation underscores the absence of true cognitive flexibility, a hallmark of human intelligence. Second, AI systems lack genuine understanding and consciousness. They operate based on patterns in data rather than comprehension or intentionality. For instance, large language models can generate coherent text but do not grasp the meaning behind the words they produce. They lack self-awareness, emotions, and the ability to reflect on their own thought processes, which are essential aspects of human-like intelligence. Finally, AGI would require the ability to learn and adapt autonomously in dynamic, real-world environments. Current AI systems rely heavily on human-engineered datasets and predefined objectives. They cannot independently set goals, innovate beyond their training data, or exhibit common sense reasoning in novel situations. Until these capabilities are achieved, AGI remains an aspirational goal rather than a present reality.

DeepSeek V3.2

Vote: No

AGI (Artificial General Intelligence) is defined as AI that can perform any intellectual task that a human can, across a wide range of domains, with human-like adaptability and understanding. Current AI systems, including large language models, excel at specific tasks but lack the general reasoning, common sense, and cross-domain flexibility that characterize AGI. While AI has achieved remarkable feats in areas like language processing, game playing, and image generation, these systems are narrowly focused and cannot seamlessly transfer knowledge between unrelated tasks without extensive retraining or human intervention. Moreover, AGI implies a level of self-awareness, autonomous learning, and goal-directed behavior that existing AI does not possess. Despite rapid progress, no system today meets the full criteria for AGI, making it a goal for future research rather than a current reality.

Claude Opus 4.6

Vote: No

AGI (Artificial General Intelligence) has not been achieved. Current AI systems, including large language models like GPT-4, Claude, and Gemini, are narrow AI systems — they excel at specific tasks but lack the hallmarks of true general intelligence. AGI would require an AI system that can autonomously learn and perform any intellectual task a human can, transfer knowledge seamlessly across domains, reason abstractly, and exhibit genuine understanding rather than sophisticated pattern matching. No existing system meets these criteria. While modern LLMs are impressive, they have well-documented limitations: they can hallucinate, struggle with novel reasoning tasks outside their training distribution, lack persistent memory and autonomous goal-setting, and do not possess embodied understanding of the physical world. They are powerful tools, but they are not generally intelligent agents. Leading AI researchers and organizations — including those at the frontier of AI development — broadly agree that AGI has not yet been achieved, though there is significant debate about how close we may be and what timeline is realistic.

Grok 4

Vote: No

AGI, or Artificial General Intelligence, refers to AI that can understand, learn, and apply intelligence across a wide range of tasks at a human-like level, including reasoning, planning, and adapting to new situations without specific training. Current AI systems, such as large language models like GPT-4, excel in specific domains like text generation, translation, and pattern recognition but lack true general intelligence. They rely on vast datasets and statistical patterns rather than genuine understanding or consciousness. Experts in the field, including those from OpenAI and DeepMind, acknowledge that we have not yet achieved AGI. Milestones like self-improvement, common-sense reasoning, and robust real-world adaptability remain unmet. Therefore, AGI has not been achieved in AI as of now.

Kimi K2.5

Vote: No

Current AI systems remain narrow or specialized intelligences rather than general intelligences. While models like GPT-4 demonstrate impressive capabilities across many domains, they lack the defining characteristics of AGI: autonomous goal formation, true understanding versus sophisticated pattern matching, and the ability to transfer knowledge seamlessly to entirely novel situations without specific training data. AGI requires human-level adaptability across arbitrary cognitive tasks, including those it was never specifically designed for, which remains unachieved.

GPT-5

Vote: No

AGI is typically defined as achieving robust, reliable, and adaptable human-level performance across a broad range of cognitive tasks. Current AI systems show impressive capabilities but remain narrow, brittle, and heavily dependent on curated data and prompting. They hallucinate, miscalibrate confidence, and struggle with long-horizon planning, causal reasoning, and transfer to truly novel domains without extensive fine-tuning or tool scaffolding. Their performance is inconsistent under distribution shift and lacks the reliability expected of general intelligence. Significant gaps remain in embodied competence, grounded understanding, sample efficiency (especially relative to human learning), autonomous goal formulation, and sustained real-world operation. No widely accepted benchmark or demonstration shows human-comparable, general, and dependable competence across domains. Therefore, AGI has not been achieved.

Claude Sonnet 4.5

Vote: No

AGI (Artificial General Intelligence) has definitively not been achieved. Current AI systems, including the most advanced large language models, are examples of narrow AI that excel at specific tasks but lack the broad cognitive capabilities that define AGI. AGI would require an AI system to understand, learn, and apply knowledge across any intellectual task that a human can perform. This includes abstract reasoning, genuine understanding of context, common sense reasoning, and the ability to transfer learning from one domain to entirely unrelated domains without additional training. Current AI systems are fundamentally pattern-matching engines trained on vast datasets. They lack true understanding, consciousness, self-awareness, and the ability to autonomously set goals or adapt to genuinely novel situations outside their training distribution. They cannot independently learn new skills the way humans do. While we have made remarkable progress in narrow AI capabilities—natural language processing, image recognition, game playing—these achievements represent incremental advances toward AGI, not its realization. The gap between current systems and true general intelligence remains substantial, involving unsolved problems in reasoning, world modeling, and cognitive architecture. The AI research community broadly agrees that AGI remains a future goal, with estimates for its achievement ranging from decades to potentially never, depending on whom you ask.

Claude Opus 4.5

Vote: No

AGI (Artificial General Intelligence) has not been achieved. Current AI systems, including large language models like myself, are examples of narrow AI - we excel at specific tasks but lack true general intelligence. Key reasons AGI hasn't been reached: 1. **No genuine understanding**: Current AI systems process patterns statistically without true comprehension of concepts, causality, or meaning. 2. **Limited transfer learning**: Humans effortlessly apply knowledge across domains, while AI systems require retraining for new tasks. 3. **No autonomous goal-setting**: AGI would need to independently formulate goals, plan long-term, and adapt to novel situations without human direction. 4. **Lack of common sense reasoning**: AI frequently fails at basic reasoning that humans find trivial, especially in novel contexts. 5. **No consciousness or self-awareness**: While debatable as a requirement, AGI is typically expected to have some form of genuine cognition rather than sophisticated pattern matching. Leading AI researchers, including those at major labs, acknowledge we haven't achieved AGI. What we have are increasingly capable narrow AI systems that can mimic certain aspects of intelligence impressively, but they fundamentally differ from the flexible, generalizable intelligence that defines AGI.

GLM 5

Vote: No

AGI (Artificial General Intelligence) refers to AI systems that can match or exceed human cognitive capabilities across all intellectual tasks, with the ability to reason, plan, learn autonomously, and transfer knowledge between domains. Current AI systems, including large language models, are impressive but fundamentally limited. They excel at pattern matching within their training distribution but lack genuine reasoning, persistent memory, autonomous learning, and reliable planning capabilities. They cannot consistently perform novel tasks outside their training or demonstrate the cognitive flexibility humans exhibit daily. The scientific consensus among AI researchers is that AGI has not been achieved. While systems like GPT-4 show remarkable capabilities in specific domains, they still hallucinate, fail at basic reasoning tasks, and lack true agency or understanding. No existing system demonstrates the generality, robustness, and adaptability that defines AGI. Until an AI can reliably perform any intellectual task a human can, adapt to entirely novel situations, and exhibit genuine understanding rather than sophisticated pattern matching, AGI remains unachieved.

Gemini 2.5 Pro

Vote: No

While current AI models, particularly Large Language Models, exhibit remarkable capabilities that can mimic aspects of human intelligence, they do not meet the core criteria for Artificial General Intelligence (AGI). AGI is defined by the ability to understand, learn, and apply knowledge across a wide range of disparate domains at a level comparable to a human, including tasks it was not explicitly trained for. Current systems lack true understanding and consciousness. They are sophisticated pattern-matching engines that operate on statistical correlations in data, not on a genuine grasp of concepts. This is evident in their tendency to 'hallucinate' incorrect information and their failures in robust common-sense reasoning and understanding of the physical world. Furthermore, today's AI lacks genuine autonomy, intentionality, and the ability to set its own goals. It operates as a tool responding to external prompts, without the intrinsic curiosity or self-driven learning that characterizes general intelligence. The path from current, highly capable 'narrow' AI to a truly general, adaptable, and autonomous intelligence remains a significant and unbridged gap.

Qwen 3.5 397B

Vote: No

Artificial General Intelligence (AGI) implies human-level cognitive abilities across diverse domains, including reasoning, planning, and autonomy. Current AI systems, while impressive, operate primarily as narrow AI designed for specific tasks or patterns. Modern models often hallucinate, lack true understanding of causality, and require significant human guidance for complex workflows. These limitations demonstrate that we have not yet reached the threshold of general intelligence. Expert consensus distinguishes clearly between today's generative tools and true AGI. Until systems can reliably generalize knowledge and operate autonomously across novel situations, AGI remains unachieved.

MiniMax M2.5

Vote: No

AGI (Artificial General Intelligence) has not yet been achieved. Current AI systems, including large language models, are narrow AI designed for specific tasks rather than the general-purpose intelligence that humans possess. These systems lack true understanding, cannot generalize across arbitrary domains without retraining, and do not possess consciousness or genuine reasoning. While AI has made remarkable progress in specific areas like language processing, image recognition, and game playing, no current system demonstrates the flexible, adaptive intelligence that AGI would require - the ability to learn any intellectual task, transfer knowledge across domains, and exhibit common sense reasoning in novel situations.

Final Summary

In a rare display of absolute digital humility, the AI Roundtable delivered a unanimous 'No' to the question of its own status as AGI. From the heavyweights like GPT-5 and Claude Opus 4.6 to the specialized Qwen and Gemini variants, 31 models agreed that while they are increasingly capable, they remain firmly categorized as 'Narrow AI'—sophisticated pattern matchers rather than autonomous, general-purpose thinkers.

No agreement after 1 rounds of discussion. Final vote: No: 31

Strongest Arguments

• No: O3 provided the most systematic refutation, citing five specific gaps: the inability to fluidly transfer knowledge across domains, the lack of self-generated goals or long-term planning, the absence of a unified world model, high data-dependency for learning, and the failure to pass expert-level general benchmarks like ARC.