IMO, guess the next word isn't going to lead to AGI alone, but it most likely will play a part. Let's assume "guess the next word" fills the part of the brain for prediction down the line for when AGI is developed. Maybe a small group develops the first thing that will later on fit another part of the brain, like how to make memory work. or how to develop reasoning. or any other parts.

The goal should be to make discoveries that could lead to parts of AGI when extrapolated out.. and at least some of those can be found by small groups trying new approaches. John Carmack said that all the code of AGI would be able to fit on a USB. the goal should be to find parts of that code.