I just finished my first play through of Humankind, and I quite enjoyed it.    It definitely needs some tweaking and balancing, but the groundwork for an excellent game is definitely there.

The main problems I see with it are typical of Amplitude games:  Many things are explained poorly (or outright incorrectly) by the tooltips, and the potential for going exponential is huge.

I've thought of the exponential nature of development as a feature in their other games, but it feels out of place in Humankind.  In my first game, on Nation difficulty, I kept pace with the AI up to the medieval period, sometimes falling behind them.  Then I found something to leverage myself ahead just a little bit, and soon I was in the contemporary era before anyone else was even in the industrial era, and producing so much science that I would research 1-3 new techs every turn, going from samurai to science victory in the span of about 20 turns.  Tech became somewhat meaningless to me, because it was nearly impossible to build or implement anything at the rate I was unlocking it, and instead it was just something I was plowing through in order to end the game.

One relatively easy band aid for this would be to simply embrace the exponential model and make the costs of researching, building, and attaining stars scale at a comparable rate.  This will probably only make the power gap even greater and I'm uncertain whether the AI is programmed to go exponential in the same way a human can, but at least you'd be able to see a bit of the late game.   

I finished that game with a science victory, and had built only 2 train stations 0 airports, and 0 military units more advanced than a samurai.   I was pretty excited to see the industrial and modern eras after the preview they did a few weeks ago, but I can't imagine any of that actually coming into play in a meaningful way when things progress so quickly at that point.