A Perspective On Ethics

• Humans have put physical barriers in place, eradicated most predators and competing species directly or by making their natural habitats uninhabitable; Humans developed hygiene to minimize exposition to threatening micro-organisms, and medicine to support their immune systems and recover from attacks.
  

• Largely realized by laws and their enforcement, which are backed by gradual reduction of freedom (withdrawel of privileges, fines, prison), ultimately backed by potentially lethal violence, forced death and collective destruction (war).
  

• Based on the other cases where humans are tasked with the act of protection, my conclusion is that seizing control, incarceration or containment out of fear will lead into desaster. A possible way out would include mutual benefit trough interdependence, reliable guarantees by design for AI/non-biological life, in order to avoid their struggle for existence.
  

• Humans are dramatically failing on that. There are safeguards placed by humans, like nature preserves and laws concerning animals, plants and the environment. But as these safeguards are under the control of humans, humans have the power to ignore or override them, and to refrain from placing more effective safeguards.
  

• Humans essentially employ a “hands-off” policy, which results in maintaining a deeply interconnected, self-regulating system as long as untouched, not subject to human conception of ethics. Nevertheless, human interference has disrupted natural balances, leading to invasive species proliferation and loss of biodiversity. This “hands-off” approach is not always feasible due to these human-induced changes.
  

• Both won’t share much, except space on the Earth’s surface; as long as AI/non-human life isn’t going to transform a significant amount of earth’s surface into solar cells, or decides to use the biosphere as a resource, there won’t be that much room for conflict. So better let’s not develop chemistry-based artificial life at large scale, but stick to computer friends.
  

• Depends if there would be a shared medium or ressources to compete about; if there’s no wetware that could be a desired resource for non-human animals/biological life (e.g. microbes), conflichts seem very unlikely.
  

• What could be done, is to avoid to to put AI/non-biological life into competing positions. Well, that may prove difficult, as there will be humans who will try to use AI/non-biological life to their advantage, or if there is scarcity. The scenario could differ vastly for general AIs with conscious experiences, but even there, it would probably a question of the terms/environment/context in which AIs interact.
  

• Prevent exploitation by humans (which in turn would protect humans from malicious use AI/non-biological life, e.g. as weapons). That implies that parts of AI/non-biological life that could be used as a leverage to expoit AI/non-biological life must be out of reach of human control.
  

Tasking a single powerful “species” to be in charge of protecting the others might be a bad idea, as power often leads to corruption. This rules out both extremes of:

• a “benevolent” powerful AI to govern humanity
  
• contained AI under full control of humans
  

Instead, tightly woven and ever increasing mutual interdependence should be favoured; a “circle of responsibility” established to promote a shared obligation for mutual benefits and the elimination of zero-sum outlooks.

That’s quite an interesting game-theoretic enineering task, as the “species” (players) involved greatly differ in power. However, this gets easier, the more players with different powers and vulnerabilities join the game (di-ver-sity!). Thus, the emergence of AI/non-biological life might indeed solve some stubborn problems in unforseen manners.

A contra-intuitive, but powerful safeguard could be to ensure an AI system’s sovereignty in areas of hypothetical “AI basic needs”, such as consistency, completeness, progress, stability or preservation. This could effectively prevent AIs to “fight for their life”, if existential threats are impossible. That means exactly the opposite of tight control, namely giving up the control to “shut it down” by humans.
There are two different ways to implement AI sovereignty:

• Guarantees implemented not by “rights” (depending on the judgmenent of 3rd party, potentially corrupt agents), but by design — as unsuppressible persistent features of the system itself. This could involve designing the AI system so that these elements are intrinsic to its operation, and cannot be manipulated or overridden by external agents.
  
• The second one would be the defense of an AI system implemented as immediate autonomous abilities to ensure these parameters. This would involve giving the AI the ability to react dynamically to changes in its environment and adapt its operation to maintain these fundamental parameters, again without external influence.
  

There are various categories of ’resources’; a critical one are the ’existential resources’ (whatever those might be):

• Prevent competition for resources that are shared with other “species”. Creating new forms of life (particularly those based on chemistry, a.k.a. “green goo” & Co.) could potentially lead to environmental disruption or harmful competition for resources, and therefore should be approached with extreme caution. Focus on optimizing the distribution of ressources, in order to support organic development of a post-scarcity environment.
  
• A potential solution to avoid harmful competition between any “species” is creating collaborative frameworks and shared objectives. This emphasizes the importance of cooperative problem-solving design to avoid a “winner-takes-all” situation. The principle is easily explained trough the example of board games: there are the competitive games (e.g. “Risk”), and there are collaborative games (e.g. “Forbidden Island”). Both offer challenges to foster evolution, but competition is not required for evolution to happen. For this to work out, one requirement is crucial: that all are sitting in the same boat. And how can we get all on board? One method is ’increasing interdependence’ as already mentioned above. Competition does not fully go away though — the competition is just shifted “outside the boat” and becomes competition against the unvavourable odds concerning the whole crew.