OpenAI's Sam Altman expresses dissatisfaction with GPT-5, stating that it doesn't meet AGI standards, despite Microsoft's partnership with the company showing signs of weakening, with Altman saying "it's got something missing."
In the rapidly evolving world of artificial intelligence (AI), the debate on Artificial General Intelligence (AGI) has taken centre stage. Notable figures in the tech industry, such as Sam Altman, CEO of OpenAI, have suggested a shift in focus from AGI to self-replication. This change in perspective comes as the tech world grapples with the complexities and resources required to achieve AGI.
Achieving AGI is no small feat, requiring vast resources, including cooling water, GPUs, AI talent, and substantial financial investment. The latest model launched by OpenAI, GPT-5, has fallen short of users' expectations due to glitches, bugs, and unresponsiveness. According to Sam Altman, GPT-5 does not meet the AGI threshold.
Despite these challenges, OpenAI, Google, and Anthropic are making significant strides towards AGI. Demis Hassabis, CEO of Google's DeepMind, claims that society is on the verge of achieving AGI. However, Hassabis warns that society may not be ready to handle the implications of AGI, and the prospects keep him up at night.
The pursuit of AGI has not been without controversy. OpenAI has faced backlash from users due to its abrupt decision to deprecate GPT-5's predecessors and bury them behind a $20 ChatGPT Plus paywall. Moreover, a separate report suggests that OpenAI could prematurely declare AGI via an AI coding agent that surpasses the capabilities of an advanced human programmer. This premature declaration could potentially sever OpenAI's ties with Microsoft, given that Microsoft's CEO, Satya Nadella, shares the same viewpoint as Altman, with a focus on delivering real-world impact using AI.
The criteria for defining AGI go beyond surpassing human cognitive capabilities. Current criteria focus on functional generality, learning from first principles, and robust performance across diverse tasks without specialization. These criteria are being formalized through benchmarks and regulatory guidelines, with emphasis on computational scale, task generality, and cognitive requirements.
The EU AI Act and its Guidelines, for instance, define functional generality and compute threshold criteria for General Purpose AI (GPAI) models. Emerging benchmarks like the Artificial General Intelligence Testbed (AGITB) provide technical, cognitive-level benchmarks emphasizing learning from first principles and generalization without prior knowledge.
In summary, the pursuit of AGI requires a broad focus on general capabilities, computational thresholds, the ability to learn and generalize from scratch, comprehensive benchmarking, and structured assessment frameworks reflecting performance and autonomy gradations. These criteria are increasingly codified in emerging regulatory frameworks and rigorous, computationally grounded benchmarks.
As the quest for AGI continues, OpenAI is reportedly ready to battle anticompetitive business behaviour in court. Insiders claim that Microsoft holding out is its nuclear option, designed to protect its massive investment in the ChatGPT maker. Sam Altman, however, promises with a high degree of scientific certainty that GPT-5 will be smarter than GPT-4.
References:
| Source | Contribution to AGI Criteria Definition | |--------|---------------------------------------| | [1] EU AI Act GPAI Guidelines (2025) | Defines functional generality and compute threshold criteria for GPAI models with lifecycle obligations | | [2] AGITB Benchmark (2025) | Provides technical, cognitive-level benchmarks emphasizing learning from first principles and generalization without prior knowledge | | [3] Wikipedia summary (2025) | Offers taxonomy of AGI performance/autonomy levels & distinction from narrow AI and superintelligence | | [5] EU AI Act Code of Practice (2025) | Details operational framework for AI providers relating to GPAI compliance |
This synthesis reflects the state of the art and regulatory consensus as of mid-2025.
Microsoft plans to invest heavily in developing software solutions for improving Windows and Office, as well as potentially expanding into the gaming industry with Xbox updates. They are likely keen on staying competitive in the technology sector and might seek synergies with OpenAI's AGI research.
In the world of finance and investing, business leaders keep an eye on the potential impact of AGI on the economy. As AGI becomes a reality, it could revolutionize sectors like finance, artificial intelligence, and technology, leading to significant changes in asset allocation and business strategies.
Sam Altman's focus on self-replication in AI research hints at a possible fusion of AI and robotic manufacturing technologies. This could have far-reaching implications for the manufacturing industry, potentially automating labor-intensive processes and boosting efficiency.
Xbox, Microsoft's gaming platform, could benefit from advancements in AI, with implications for game development, user experience, and even for-profit ventures such as in-game advertising. By leveraging AI, Microsoft may enhance the gaming experience for millions of users worldwide.
With AGI comes the potential for cutting-edge advancements in artificial intelligence and technology. This could lead to significant growth opportunities for software companies like Microsoft, which could capitalize on these advancements to create new, innovative products and services.
