The undisclosed feature of GPT-5: it makes decisions on your behalf, and you may not always be aware of it
The latest artificial intelligence from OpenAI brings a significant change: with GPT-5, users no longer need to choose which model to use, as the AI will make the decision for them. While this simplifies the process and reduces confusion over names and versions, it also means users lose the ability to decide unless they opt for a paid subscription. This balance between simplicity and control may not be as straightforward as it appears.
### A model that is many models
GPT-5 is a combination of various variants, each with different strengths such as speed, power, and specialized reasoning abilities. The innovation lies in the internal router that automatically selects the most suitable variant based on the conversation’s complexity, context, or required tools. Users can still request a specific type of response by clearly stating it in their message, but overall control lies with GPT-5.
### The hidden side of simplification
While this approach simplifies the user experience and makes AI more accessible to beginners, it also introduces opacity. Not knowing which variant is being used may not bother some users, but for advanced users, it means losing the ability to choose the most appropriate tool for each task. This excessive simplification could lead to a loss of transparency and limit users’ ability to make informed decisions.
### The privilege of paying to decide
For a monthly fee of $200, subscribers can manually activate any internal variant of GPT-5, retaining full control over the AI model. Non-subscribers, on the other hand, rely on the automatic router for decision-making. Previous models like GPT-4.1, GPT-4.5, and others are discontinued with the introduction of GPT-5, except for those with existing subscriptions.
While a simple solution would be for GPT-5 to disclose the version used in each interaction for transparency and future adjustments, this feature has not been openly implemented by the company. In this new landscape, average users benefit from convenience, but experts may lose visibility and control over their interactions with AI.
