There are some good arguments in favor of understanding corporations as artificial, intelligent meta-(or super-)organisms:
- Corporations are aware of their predators (government agencies) and aim to extinguish them (see regulatory capture)
- Corporations prioritize their own survival over secondary goals (profits, growth, etc). To avoid death, it may shed massive amounts of its body (close stores, fire employees etc), always with the hope of bouncing back
- Corporations can reproduce. For example, while the F.W. Woolworth corporation no longer exists in the US and UK, one of it’s children, the German division, still operates. In fact, this child corporation is now is expanding to more countries.
- The “neural network” inside the corporation (the employees) may be relatable at a personal level and each “neuron” may understand what’s around them, but assuming a large enough corporation , there generally is nobody who knows everything, and the overall behavior of the corporation is difficult to make sense of.
- Corporations have an “immune system” (human resources). HR is concerned with keeping the overall corporation body “healthy” by trying to retain “good cells” (productive employees) while disposing of “sick” cells (employees which may pose a risk to the corporation).
- Sometimes, corporations develop cancers: parasitic employees (often CEOs) which as their primary goal have to strip as much of the resources of the corporation for personal gain are sometimes cancerous in nature. The corporation’s “immune system” in the form of HR or a board of directors is supposed to prevent this from happening, but given a cunning enough CEO, it may be powerless and even reward destructive behavior (such as share buybacks or short-term greed).
- Corporations tend to be very happy to screw over their environment (eg. polluting rivers to the point they catch fire), their competitors, and sometimes even their partners if they intend to devour (buy) them instead of merely paying them.
The idea that corporations can be understood as an AI isn’t new – I got introduced to it by Charles Stross in his “Dude, you broke the future!” talk. They are slower to act than computer-based AI, but in return, for better and for worse, there’s virtually nothing they cannot do, as their base “neurons” (people) are each natural general intelligence agents (“NGI agents”). The use of NGI agents as their base unit makes the entire corporation AGI, though notably not identical to it. A corporation will never experience love like humans, though it certainly can be dazzled and allured by certain things and trends. However, a corporation will have experts in many fields and roles, so the overall institutional knowledge is far greater than any employee within can carry.
Corporations both are used in an agentic AI role and use agentic AI: A third party may contract them to do something for them (like: “build me a yacht covered in gold”), and the corporation itself may contract out individual tasks to agent AIs, which may be either other corporations, groups within the company, or people (NGIs). NB: Often, groups within a corporation (departments, divisions, etc) can be considered AI themselves, as every department will struggle for power and resources within the overall corporate system.
With all of these things in mind, here’s an exercise for the reader, and the first time in a very long time where I’ll open up the comment section for discussion:
- What does AI safety look like for corporations?
- Where did corporate AGI cause harm to humans?
- What are humanity’s goals?
NB, in principle, countries, cities and some other groups of humans may be understood as AI as well, though not all groups formed by humans are (or act as) AI. - How can we get corporate AGI aligned with humanity’s goals?
- On a broader picture, are there any AI systems which are especially misaligned with humanity’s goal?
- Can we apply any of our findings to computer-based AI? Why or why not?
I’ll try working this out myself as well, and may publish a followup to this down the road. Looking forward to your thoughts!
Leo Wattenberg 