The Scaling Hypothesis and Future of AI Models

The current wave of AI technology is driven by the scaling hypothesis, which suggests that AI systems improve with more data and computing resources. However, by 2024, the cost of building and operating large AI models is becoming a significant challenge. For example, the most advanced AI system, GPT-4, costs around $150 million to build and operate the data centers. The next system after GPT-4 could cost around a billion or more. The question arises of whether there will be enough funding for building these advanced AI models, and if the scaling hypothesis will still hold true when more computing power and data are used.