AI Tech expenditure

[Parkerrush92]

Hi all,

AI upgrade being requested.

I get tech offers for a lot of unrelated products, that I have technology of only say 25.

The AI spends excessively on tech for products that are not the focus of their companies.
That being said, additional improvements should be made to focus the ai on products that make them the most money.


For example:

AI company makes 35 different products.
5 of these products net more than 200 million in profit

Instead of spending on tech and production for a new low necessity indexed product, ai should focus on improving existing margins and profits, and advertise for market share.

I'm happy to be of assistance.

Parker

[anjali]

sorry to pull up a post thats nearly 2 month old but the A.I.'s just getting more brainstorms day by day and i havent found newer posts to this topic but it seems that the more power you give them n the easier ya make it for them (pump up with expertise, cash, limit products, all ressources availible at seaport) the more nuts they get.

setup:
on basis of realworld mod but altered
10 cities each 5 million residents
1 seapot each city (consumer + industry) quality=low, constant supply=no

5 product classes only
Fashion (leather goods,Apparel,Footwear)
Jewelry
Watch
socks, wool sweater, Silver watch, Leather Wallet, Gold ring given at start to everyone with standard tech 30, everything else to be invented

40 companies, very aggressive, very high start capital
35 diversified,
5 retailers,

expertise set to 40 companies (incase a retailer switches to diversified)
20 Apparel,
15 jewelry,
5 watch,

with other words, about $35bn against my Private company with 5 million (no financial actions nor stock market)

the first 3 years inventions were made, then heavy tech trading started, and factory were put up 1 factory per product per company. after this not a single new invention only tech pumping on existing and permanent tech trading therefore i got 40 A.I.'s against me with tech 200-350, brands of 80-95 i should'nt be making a dime

invented in 10 years total
Glas,
Steel,
all jewelry exept sunglasses
not a single leather good, footwear nor apparel (20 A.I.'s are apparel specialists !!!)
and only jewelries are produced a shock with 20 Apparel expertise A.I.s

now the funny part ...
most A.I.'s got up to 25 shops for their 1 to 6 factories ...25??? not to mention that there are 5 retailers comming up with another 30 shops. thats a total of 37 factories vs 152 shops (it were much more shops but in the last roughly 6 month 45 shops closed and more then half of all existing R&D's due to this stupid stuff they are doing as explained below

factories pulling $25 million on a single A.I. per year and the 25 shops ./.$150.000.000!!!
they could set up 50 more factories and wouldnt be able to cover demand but they only put up more department stores which cant even get supplied because of the totally over stretched factories and more R&D's get more expanses to go faster bancrupt
spending on advertisning is maxed out on every single sales slot they have thats where the loss is from, without they would make profit. they run it until they are maxed on loans and $0 in the pocket

then they start to rip down 3-10 shops in one go to stay afloat
next month, they put up a new R&D straight away closed down because they didnt set it upwith their last bucks in the pocket and had to rip it down as they instantly run into red numbers and another 2 shops closed
month 2: new R&D up(or 2 R&D's if they got the money for 2 to make sure they go into red numbers) / instant down as out of cash and another 2 - 3 shops closed to stay afloat
this continues until the company value is around $10-30 million then the factories pull enough profit to support 0 R&D's and 5 shops with full advertise spending
now they put up more shops and R&D's again without bothering the 150 million loans and 3 month later we start the same game again ... close down, open up 3 close down, 1 open up till bancrupt.

just got it again, now the biggest guy left in the game 6 factories 25 shops .... january 18th closed down R&D (out of cash) and february 20th he had to close his last R&D, 2 days later with his last pennies open up new R&D and instantly closed down du to out of cash. he pulls 50 million loss a year with 6 factories on heavy profit and 25 shops on heavy loss and instead of consolidation he keeps destroying its company assets. In real life he d go to prison for 10 years minimum for this bullshit as Head of a public body (company on shares, ltd whatever you name it)

and its not 1 A.I. ... ALL of them doing this stupid shit. Capitalism plus A.I. out of 1995 was thousand times more intelligent to be honest. i would have trouble to survive this with only 9 companies against me (i.e. "staying in vogue" scenario, thats what i m currently working on, to get this setup in capitalism Lab, but really those A.I.s in cap lab are no match, the only difficulty on my current setup is that i ve no cash to put up R&D's nor lots of shops/factories no wonder by 5 million starup with no loans)

appologies if that sounds angry, but i am really annoyed about this brainstormed A.I.s in cap lab

[David]

Have you tried a normal game where the AI's focuses are not scripted and restricted to specific product classes ?

I just tested a game and my subsidiaries are having decent performance, as shown in the screenshot attached.

I think it is important to recognize that the intended design goal of this DLC is to ultimately let the player have fun playing it. Imagine the case where your subsidiary's AI is always capable of achieving excellence, then all the player has to do is just to maintain control over the subsidiaries. This kind of strategy will be too linear and inevitably boring.

So from the perspective of AI programming, it is true that there is always room for improvement. But from the perspective of game design, the real question is whether it is our design goal to make a game with the best possible AI and then let the player control of the companies of these AI and automatically beat the game?

We believe that even if we did improve the AI to a level that most players found competent, it may not necessarily improve the gameplay. It is just like, in a RPG game, giving the player character with all attributes max out at the beginning of the game, which may sound nice but in practice the player will be bored soon after realizing that every monster in the game can be beaten effortlessly.

[anjali]

before i used the realworld mod, i did play the non dlc 2.7.x but it got me always totally bored after a few years of ingame time as i did sit on emdless cash and the A.I. were just no match at highest levels set. also the picture / logos i disliked in cap 2/cap plus and was missing the alternative / food+beverage product lines which i had in cap plus. thats why i started to try the beta with DLC and realworld mod, but the whole realworld was too much for the A.I.'s and i got the same result of as you can see in my saves which i uploaded for you in the stock issue post. 6 billion assets in less then 4 years and hundreds of millions every month availible. thats the reason why i did try to limit it heavily to make the game demanding and in my current set up i m sitting after 10 years on just 6 shops (2x2 footwear there was no way to set up a department store at the start of the game with just 5 million in the pocket even in cheapest town its 5.3 million + monthly without a single sales unit) 3 small factories and a single large farm not a single r&d and with 600k$ in the pocket and 35 million total profit across 10 years. once i run completely through this 100 years i gonna try again a regular game and monitor it. but regular is just way toooo easy.

of course i am not looking for a super A.I. but some form of cost management must be there, else a single subsidary you set up will cost ya either billions of losses a year or 1000 times more time to correct every 5 minutes its bullshit and honestly even a 3-6 million monthly profit is just not enough (inflation inverse) as they keep sitting even without mod on hundreds of millions cash (their wallet reduces alone by inflation 35-70 million a year) ofcourse if i check the corporation overview, the graphs looking great for a few years on them (30 years graph shpws some comapanies spiking up to 150 million profit and then rouhly half of the corps drop into 200-400 million loss in the graphs as they suddenly all start to have 2 to 3 times more expanses then revenue)

[David]

before i used the realworld mod, i did play the non dlc 2.7.x but it got me always totally bored after a few years of ingame time as i did sit on emdless cash and the A.I. were just no match at highest levels set.

I have recommended to you several ways to set up the game to make it more challenging in another post, which you may read at: http://www.capitalismlab.com/forum/view ... 813#p13813