Humankind is Awesome :) , but too easy :(

The AI certainly doesn’t catch up, so if you are ahead that’s pretty much the game.

That is why I want to see increasing AI advantages by era instead of straight line like it is now (and is in most games). You start behind, you catch up, you get ahead. I'd mentioned in other threads the following:

HK level (or new Genetically Enhanced level)
Combat Bonus = +2 * (era + 1), so from +2 in Neo to +14 in Contemporary (linear increase)
Worker Bonus = +3 * (era*era), so from +3 in Ancient to +108 in Contemporary (exponential increase)

I believe this would allow the AI to remain even with the human player. Of course, some of their other modifications may do just that. Wish all games would feed into a Deepmind like AI so it would just figure out how the human wins and continue to learn and modify, eventually having to dumb it down to have a chance to win. Making AI play well is VERY hard.
 
That is why I want to see increasing AI advantages by era instead of straight line like it is now (and is in most games). You start behind, you catch up, you get ahead. I'd mentioned in other threads the following:

HK level (or new Genetically Enhanced level)
Combat Bonus = +2 * (era + 1), so from +2 in Neo to +14 in Contemporary (linear increase)
Worker Bonus = +3 * (era*era), so from +3 in Ancient to +108 in Contemporary (exponential increase)

I believe this would allow the AI to remain even with the human player. Of course, some of their other modifications may do just that. Wish all games would feed into a Deepmind like AI so it would just figure out how the human wins and continue to learn and modify, eventually having to dumb it down to have a chance to win. Making AI play well is VERY hard.

I reckon the bonuses scaling with era is the key, they get a lot of initial bonuses but need a few nudges to help them scale exponentially along with the player. When you look at the graphs you can see the leading AI sometimes starting to get onto an exponential path, they're usually well behind in Industry though which is the main driver of growth.

Pretty much everything that's called "AI" is just a series of nested "if..then" statements, so making the computer play well is a lot of work. And if you can make an algorithm competitive at a game, it'll become trivial for human players and very boring. That's why non-symmetry is almost essential in some form for games to be interesting, if the computer can't cheat it won't be a fun game.

The amount of effort that gets put into real AI like Deep Mind is huge and well beyond the scope of any retail game.
 
The amount of effort that gets put into real AI like Deep Mind is huge and well beyond the scope of any retail game.

Wish there was a Bill Gates of gaming that would literally waste billions of dollars doing this. I can dream.
 
In terms of fame, I'm finding that by picking random cultures I'm not rushing into each new era, because most EQs aren't that powerful, and there's no risk of "missing out" on a culture because I'm just randomly rolling whatever's left - so there's no reason to move forward other than tech. This means I'm getting a load of fame and I'm thousands of points ahead of the AI from Medieval. My feeling is that the "behind on fame and then catch up at the end" dynamic only really applies when you're aggressively chasing culture pick and EQs, so it could definitely be made harder.

Wish there was a Bill Gates of gaming that would literally waste billions of dollars doing this. I can dream.

Spoiler :
 
Great discussion, but keep in mind I'm not asking for a mastermind AI at this stage or taking HK AI to Civ 5 VP level. It's not like I mastered the game after playing for months/years. All I'm asking for is to have the AI pose a challenge at the highest difficulty having played only 5-6 partial games by simply pursuing a fame collection strategy (which is the whole point of the game.)

Another interesting fact: I posted virtually the same thing on the official HK forum last night but not a single soul has responded yet, in contrast to this forum lol
 
You guys are both right. The AI does catch up sometimes but sometimes it may seem like it will catch up but it doesn't. I've had a 1st place and 3rd place because of this.

I also have had more fame than the AI but had the AI would be ahead by a star or two, or a whole era by a star or two.
 
That's reasuring to hear, in that case I will play out a game to completion even if I'm thousands ahead in fame just to see if the AI catches up. Let's see if the latest beta patch improves the AI, they just released a massive one
 
AI is farming Fame more with the beta, not rushing into the next era anymore... let's see how this translates into the end game.
 
Not sure if this game will count, starting Egypt > Persia and all, but on my 4-player continent (can we have 10 players without fear of not getting civics now?) I have a dismal 800 fame to everyone else’s 2-3k and there have been more at-war relations than peaceful ones since the moment everyone was acquainted. About to pump out 8 immortals with an eye on a city with 2 coffee and 3 gemstones.
 
That's reasuring to hear, in that case I will play out a game to completion even if I'm thousands ahead in fame just to see if the AI catches up. Let's see if the latest beta patch improves the AI, they just released a massive one

In my experience the catching up might happen if an a.i. declares a war and captures another a.i.'s lands and shots to no.1.position.

But if you check the leading a.i. and win it in a war, there is no rebounding for it and then you just snowball ahead.

The a.i. is very dumb and isn't doing some genious moves, it can sustain itself because it doesnt suffer from unit upkeep and doesnt need food and stability like how player does.
Without those bonuses it would not hold its empire for at all.
 
I don't know if it's the new patch or just because I had a very weak start but I'm getting my ass kicked in my latest game. Went into Ancient on T11 (for challenge - I usually aim for T10) with very average outpost positions and by T80 I have barely got anything off the ground.

Playing a crowded start (8 on large 3 continents with 50% land and new world) and was simultaneously attacked early by Hittites and Huns. Had to surrender to the Huns (1700 gold!) to be able to beat the Hittites, then I had to surrender to the Huns again and give a territory away before I got elephant archers. Completely boxed in by hostile neighbours at this point so I rolled the dice and started a war with the Huns. They had been fighting the Hittite/Persians for a while and I could see some epic battles flashing in the fog so I thought they might have lost a couple of their stacks. Only had to beat 2 stacks and then taking his cities was straightforward, which netted me a city, 2 territories, and 2 wonders. I'm now spread out along the very average-yielding coastline, still have basically no infrastructure up, dead last in fame, stability is a major limiting factor because I have no trade, and I'm limping towards Medieval. The only way out I think is to get rid of one more of my tyrannical neighbours so I'm going to press my elephant archer advantage into the Greeks and hope he doesn't get too many crossbows before me.

I think early tile yields make the biggest difference of anything, river forests with waterfalls mean you can start snowballing right from the start. Average tiles and suddenly everything is a struggle. I love how you can survive by mortgaging your empire though even if you're totally outnumbered and outgunned. It's much more fun fighting back from behind and this game lets you do it. It's similar to coming from behind in Civ4 but 1) there's multiple paths to catching up, not just "Cuir rush", and 2) in Civ4 you're in danger of an AI deciding they just don't like you and rolling "take over all your cities" - here they're happy with a territory or 2, or a bunch of gold, so it's not game over.

From the graphs 2 of the AI (on other continent) are snowballing hard in early Medieval, including in Industry... so hopefully the new patch is doing something good for their competitiveness.
 
The a.i. is very dumb and isn't doing some genious moves, it can sustain itself because it doesnt suffer from unit upkeep and doesnt need food and stability like how player does.
Without those bonuses it would not hold its empire for at all.

Not dumber than that other game to which we have been used to in the last 5 years, and in fact, much better. Genius? Of course not. Needs bonuses? Welcome to 4X gaming. But worse than the non-existent civ 6 AI? In another planet, maybe.

And 1 month in, already much much better than at launch. Promising, as opposed to the other lost cause.

Not sure if this game will count, starting Egypt > Persia and all, but on my 4-player continent (can we have 10 players without fear of not getting civics now?) I have a dismal 800 fame to everyone else’s 2-3k and there have been more at-war relations than peaceful ones since the moment everyone was acquainted. About to pump out 8 immortals with an eye on a city with 2 coffee and 3 gemstones.

Haha, told ya.

Did you notice that such lead comes from them staying in each era longer than the 7 stars? Watch the leaderboard every other turn or so. They are farming now, like a human does.

II love how you can survive by mortgaging your empire though even if you're totally outnumbered and outgunned. It's much more fun fighting back from behind and this game lets you do it. It's similar to coming from behind in Civ4 but 1) there's multiple paths to catching up, not just "Cuir rush", and 2) in Civ4 you're in danger of an AI deciding they just don't like you and rolling "take over all your cities" - here they're happy with a territory or 2, or a bunch of gold, so it's not game over.

Thank the War Support system for that; it's showing more and more to be a superior system, and better designed than initially thought by the masses. Can it be improved? Sure! Is it working as intended? Sure squared!
 
Haha, told ya.

Did you notice that such lead comes from them staying in each era longer than the 7 stars? Watch the leaderboard every other turn or so. They are farming now, like a human does.

I accidentally used default personas in the beta (not sure if expert play any different than normal), but while they got tons of fame in ancient, I don’t know if I saw farming in Classical. Just capturing up to 7 cities as Persia caught me up to the AI leader who was on another continent. It seems that conquest definitely triggers expansion stars now (maybe it always did but I never seemed to get them before).

I usually look for AI sitting at 7 stars, which the leaders did not do in classical this game. I stopped when I got ahead.
 
Something is a little off with today's patch. It is true the AI is now farming era stars instead of immediately going to the next era. I was expecting large amounts of AI era score but on this Huge HK difficulty map I'm at T119 and #1 in fame with 5370!?! I've pretty much been at war the whole time with my neighbors to the west and south. I did something I never did before and stuck with Greece for two eras, I might even go with them again just because it has helped me stay ahead of the tech curve. I got several halberds including three 3-star halberds just making mincemeat out of the Mauryan elephants and now I'm one turn away from Flintlock and thankfully there are two saltpeter mines that I will be able to work soon.

Thank god for two peaceful trading partners but their luxes are a tad expensive. But now that I have patronage and a few lux manufactories my stability is getting better. Also weird that 6/4 cities and I'm still getting +280 influence per turn, I'm only losing 120 per turn due to city cap. EDIT: Oops I was actually one turn away from Supply Lines and now I'm at 6/5 cities and 4 turns away from Flintlock.

I've been so busy with fighting that I've barely met anyone else. I got 4 more civs to meet!
 

Attachments

  • Humankind 2021-09-17 17-45-09-50.png
    Humankind 2021-09-17 17-45-09-50.png
    2.5 MB · Views: 33
Last edited:
Ok finally finished the Huge map game on HK difficulty and this patch has good and bad issues. The AI farming stars scared me at first but once I caught up during the medieval period I only fell behind once then I never looked back. When I was done I had 14.7k fame and the 2nd place AI had 8400. I exited French a bit early because I didn't see any oil on the map to build the science projects with Aussie so went Turks instead. I did finally get oil from far away trade partners so started on the science projects. However after building all three I didn't see the dialog to end the game :( So instead I had to research all techs so it was good I took Turks in the end.

Prior to this patch whenever I played Huge maps I always got horsehocky for resources and the odds for civics not working seemed a lot higher than normal maps so I gave up. This map there was a ton of great luxes and I finally got the There Can Only Be One achievement. Also got an MVP badge for 25 consecutive kills by a unit lol. Going Greece two eras in a row worked out as it allowed me to have the tech advantage when it was most needed. Especially halberdiers vs those Mauryan elephants. When I had musketeers they still had bowmen and pikemen.

I grabbed so much land from wars with my west neighbor that I had 45 territories in the end.

Good things:
* Huge maps have lots of lux resources(but still seem strategic strained)
* Civics don't seemed bugged anymore

Bad things:
* The 3 science techs didn't trigger a win :(
* AI doesn't seem to farm era stars very well even on HK difficulty.
 

Attachments

  • Humankind 2021-09-17 23-14-16-83.png
    Humankind 2021-09-17 23-14-16-83.png
    2.3 MB · Views: 38
Last edited:
Pretty low AI fame that game for sure. I wonder if others are seeing this, but it seems that an AI is only competitive if it starts to absorb its neighbors. Then I usually see at least one at 13-14k by this turn. This might have just been a peaceful game for AI.
 
But worse than the non-existent civ 6 AI? In another planet, maybe.

And 1 month in, already much much better than at launch. Promising, as opposed to the other lost cause.

Hmm I disagree with that, Civ a.i. has to deal with stuff like loyalty and food and unit upkeep, but Humankind a.i. just rides on without feeding its population or caring about paying troops or stability. Of course without the bonuses a.i.s would be worse both games.

Humankind difficulty

-50% time to convert territories to religion / culture (rounded up)
-70% population food consumption
+25% science
+10 industry per population
+30 industry, money and food
+30 stability
-80% unit upkeep
-60% buyout cost
+10 fortification
+2 combat strength on units

What HK does well is stacks, Civ 7 must have them back as it makes moving troops much easier for a.i.
 
Had a very tough start but once I fought my way off my continent it was easy to catch up in Fame. Admittedly I was conquering/colonizing the new world with 2 militaristic cultures in Early Modern/Industrial (Poles/Germans) for some easy militarist fame bonuses.

There's not a huge amount of game left (about 4 empires heading for a science finish) and the best AI is only just over 9000.

Spoiler :
 
Halfway through a game with the new patch and having a bit of a tough time at the moment, the AI definately seems to have improved a little.

In some battles I've actually seen the last surviving damaged AI units appearing to refuse to attack and fall back, instead of their usual kamikazee final assaults.
 
Last edited:
Top Bottom