In AI, an "agent" is an autonomous entity which acts upon its environment in order to achieve goals. In this sense, is a company an agent? Or in other words, does it make sense to think of the company as having its own goals independent of the people running the company? If so, what organizational structures need to be in place for such an agent with independant goals to emerge? Is being an agent enough to consider a company a "person", or would something more be required, like consciousness? I'll put a poll here, but really I'm more interested in people's comments.
Anything which is capable of interacting with an environment i.e. being in a state, act and optimize for rewards is an agent. In that sense, company is an agent as well where it's interacting with environment to optimize rewards (improve its business metrics). But internally, it has multiple hierarchical agents at play who have their own environments that they are learning to optimize for their rewards. Company is the environment for such internal agents and its not hidden. It's partially observable. That's why internal agents are pretty good at navigating the search space to optimize rewards. However, real world environment is more hidden in nature. That's why company as an agent has a tougher job at hand.
I think I agree about it being an agent composed of other agents. But I wonder how the company's goals emerge from the composing agent's goals. For example, say action x maximizes profit and action y is good according to some other human value (helps people, helps the environment etc.). Companies will almost always choose x. Is this still the case even if all the decision makers prefer y? For example, maybe the CEO and board are all thinking, "well I'd like to do y, but this is a company so we have to maximize shareholder value..." If this sort of thing happens then the company's goals are independent of the leaders' goals. In some sense, the company is an agent implemented on top of (or emulated by) the leader's minds.
Ya I agree. Company as an agent has its hands tied. It can't act on its own as of now. It has to be manually driven by a set of agents (currently humans) that drive it for not necessarily the maximization of the company's rewards but their own short and long term rewards. Company tries to align the rewards of the agents that control it to match its own rewards its trying to achieve. And it tries to create a conducive environment for CEO and employees to take actions that best transitions itself into a state that is nearest to its goals. But if company could be independent and not depend on these human actors, my hypothesis is it would perform more efficient. Because each human is added complexity into the system.
Very important question. Someone support me on this but the basis of this concept dates back to early days in the USA when a few lawyers argued that their clients company should have the same rights as an individual human, and that set a precedent and foundation of corporate America.
Personally I don't think companies should have the same rights as individuals
Similar to how we are a collection of cells that could possibly survive on their own, companies are pseudo life forms composed of otherwise Independent life forms that found symbiosis through cooperation
You are a true nerd
Thank you