[PRIMO Tech-a-Break] Status Check: How Close is GPT-5 to AGI?

[PRIMO Tech-a-Break] Status Check: How Close is GPT-5 to AGI?

Hello! Yesterday, you might have seen the news about the launch of GPT-5. Many of you have probably tried it yourself by now.

OpenAI has previously hinted at the arrival of AGI (if you’re not sure what this is, feel free to ask in the comments). Let’s take a moment to check how close we are to AGI. This new version of GPT-5 shows behaviors that we once thought could only be achieved by AGI. For example, it can handle a variety of tasks and choose problem-solving strategies on its own without requiring step-by-step instructions from humans.

However, it’s still not AGI – it can’t set goals on its own yet and still requires human input to provide tasks. To put it simply, it’s like “a world-class athlete who still needs a coach” as opposed to AGI, which would be able to plan and compete independently.

Let’s update you on what’s new with GPT-5.

From a Question-Answer Model to a Decision-Making System

GPT-5 isn’t just one model, it’s a system designed with three levels that work together

  1. Base Model – Quick responses, low resource usage, suitable for general tasks.

  2. GPT-5 Thinking – Uses multi-step reasoning for complex tasks.

  3. Router – Decides which model to use based on the task in real-time.ฃ

Major Changes I’ve Noticed

  • Higher Accuracy – Reduces hallucination chances and shows confidence levels in responses.

  • Larger Context – Able to manage massive documents or datasets.

  • Personalization – Can adjust its personality and response style (I haven’t tested this yet, but OpenAI claims it works).

  • Integration – Connects directly with Gmail, Calendar, Microsoft 365, GitHub Copilot, and Azure. This shows how OpenAI is expanding its ecosystem to compete with Google and broaden its market to enterprise users.

Engineering Perspectiv

From what I’ve seen, having the router between the models allows GPT-5 to use resources only when necessary for simple tasks and fully utilize its power for more complex ones. The result is improved latency while maintaining strong reasoning capabilities (though we’ll need to monitor this over time).

This is a clear step forward in AI transitioning from a “response system” to a “thinking system.” But, as for whether it’s close to AGI, I’d have to say: not yet.

Understand AI Intent Before Responding to Users to Avoid Misunderstanding >> Read more