The thing holding up AGI is data

This is an idea I have for a long time. I'm sure many know already, but for those who are new to the topic here is my opinion:<br><br>AGI would have been here now, if there was enough data for it. It's not the quality of the models themselves that need to improve, it's the lack of data that has become the main hindrance for the evolution of AGI, especially that can compete or surpass human intelligence. <br><br>There are several reasons why data is the biggest hindrance, the main being that as the current models become more and more complex and advanced, they need even more data to be trained. The need for data grows exponentially, and the data itself is becoming the bottleneck because we just don't have enough to advance the models enough before they plateau in their performance because of the lack of data.<br><br>The need for data is so acute that companies like Google and OpenAI are now aiming to create models that are as efficient as possible with data usage, and many of them gravitate towards zero-shot and One-shot models, because training data is becoming more and more expensive to create and use. We barely scratch the surface of creating a human level AGI, and the need for data is already impossibly high for what we know is needed for a model to achieve true AGI.<br><br>I think this is also not talked about enough, because as I said above, the public opinion is that the main problems for AGI are the quality and complexity of the models, which is partially true, but I think the data problem is much more important. I realize I might be wrong, and I am open to opinions that contradict or support my point of view.<br><br>&#x200B;<br><br>EDIT: It seems like many of you, if not most, disagree with my opinion. This is good. I am happy that the community is so active, and I will keep in mind the criticism and points you have raised. The post is just something I wanted to throw in and it generated a great discussion.