OpenAI has released the newest version of its artificial intelligence chatbot tech, GPT-5, which it says provides PhD-level knowledge in fields including coding and academic writing. CEO Sam Altman termed the release as an era of a new ChatGPT, one that is smarter, faster, and more reliable when compared to its predecessors.
Altman described the evolution of the model as talking to increasingly more educated people: GPT-3 was like a high school student, GPT-4 like a college graduate, and GPT-5 is like being able to talk to an expert on any topic. The upgrade, according to OpenAI, greatly decreases the number of hallucinations, or times when the AI makes up information, and exhibits better reasoning, as it shows how it works, its logic, and its conclusions.
The company is also selling GPT-5 as an all-purpose coding helper, which can generate complete software packages. It comes off the back of a wider industry trend towards specialized AI tools, with the Claude Code developed by Anthropic and the Grok developed by Elon Musk also operating in the same field.
Not all people are persuaded, however. Prof. Carissa Vliz, a scientist at Oxford Institute of Ethics in AI, warned that although such systems are impressive, they mostly imitate human reasoning and do not actually replicate it, and the hype could run ahead of the reality. Experts in ethics, including Gaia Marcus of the Ada Lovelace Institute, have pointed out that the necessity of adequate governance of AI grows in parallel with its capabilities.
Other industry players like Grant Farhall of Getty Images have raised concern over how AI models are trained and how content creators must be fairly paid. They claim that authenticity should be a priority because the AI-generated content is getting more and more convincing.
The release of GPT-5 is also accompanied by the tensions with its rival, Anthropic, which denied OpenAI access to its API, claiming a violation of terms of service. OpenAI defended its actions as industry testing protocols.
In addition to the technical improvements, OpenAI is also adding features that will encourage healthy interaction with ChatGPT. The system will not provide absolute responses to questions about personal life but rather will help users make reflective decisions.
It is still unclear whether GPT-5 will match its billing of a PhD-level model, but the release is the latest milestone in the AI arms race, which is seeing capability, ethics, and regulation colliding at an unprecedented pace.
