Why AI Singularity May Be Nearer Than We Think We are already in the age of AI improving AI. OpenAI has just developed CriticGPT, an AI model that helps identify errors in programming code generated by ChatGPT. Based on GPT-4, CriticGPT is designed to assist human reviewers in detecting mistakes and flaws. The model outperformed humans in error detection, catching around 85% of bugs compared to 25% by their peers in flesh. OpenAI says CriticGPT not only excels in code review, but also shows potential in identifying errors in non-code tasks of ChatGPT. Until now ChatGPT was refined in part by something called Reinforcement Learning from Human Feedback (RLHF). For instance, the thumbs up/down buttons you see allow users to provide direct feedback on the model's responses. The cumulative user feedback is then used to fine-tune and improve the ChatGPT model over time. Now, imagine an AI model, that's very good at catching errors and issues, doing the same and fine tuning ChatGPT in specific tasks? That's what CriticGPT is going to do from now as OpenAI plans to incorporate it into the RLHF pipeline of refining ChatGPT. This may lead to an accelerated improvement in ChatGPT's coding skills. And as similar models are created to find errors and refine outputs of ChatGPT in other tasks, we can imagine we are headed for Artificial General Intelligence (AGI) sooner than later. I am predicting that sometime in the 2030s, we will have AGI, provided governments and corporates do not stifle the current level of AI innovation and advancement. Sometime in 2030s may also be the point at which artificial intelligence surpasses human intelligence, allowing AI to self-replicate and improve itself autonomously without human assistance. That will be the start of AI Singularity. As self-improving AGI models help innovate and improve technology, we may very well see Singularity by the 2040s, as predicted by Ray Kurzweil. However, I don't believe such a Singularity will happen. The inherent nature of human societies means that powerful elites will likely engage in conflict when they perceive that other elites are gaining an advantage through AI, potentially leading to a war that could derail the entire Singularity process as it starts. Now, don't imagine countries throwing nukes at each other. The war will most likely be conflicts between powers that affects economies, production, and disrupts trade of resources needed to keep AGI moving forward. So I think, or in fact I am sure, the world will end up in a disabling and debilitating war of civilizations before we get to Singularity whenever it starts to happen. But wait, isn't that in the nature of Singularity too? Singularity - a hypothetical future point in time when technological growth becomes so rapid that it leads to unforeseeable changes in human civilization.