Toward Beneficial Human-Level AI… and Beyond

AAAI Conferences

This paper considers ethical, philosophical, and technical topics related to achieving beneficial human-level AI and superintelligence. Human-level AI need not be human-identical: The concept of self-preservation could be quite different for a human-level AI, and an AI system could be willing to sacrifice itself to save human life. Artificial consciousness need not be equivalent to human consciousness, and there need not be an ethical problem in switching off a purely symbolic artificial consciousness. The possibility of achieving superintelligence is discussed, including potential for ‘conceptual gulfs’ with humans, which may be bridged. Completeness conjectures are given for the ‘TalaMind’ approach to emulate human intelligence, and for the ability of human intelligence to understand the universe. The possibility and nature of strong vs. weak superintelligence are discussed. Two paths to superintelligence are described: The first path could be catastrophically harmful to humanity and life in general, perhaps leading to extinction events. The second path should improve our ability to achieve beneficial superintelligence. Human-level AI and superintelligence may be necessary for the survival and prosperity of humanity.

Formalizing Convergent Instrumental Goals

AAAI Conferences

Omohundro has argued that sufficiently advanced AI systems of any design would, by default, have incentives to pursue a number of instrumentally useful subgoals, such as acquiring more computing power and amassing many resources. Omohundro refers to these as “basic AI drives,” and he, along with Bostrom and others, has argued that this means great care must be taken when designing powerful autonomous systems, because even if they have harmless goals, the side effects of pursuing those goals may be quite harmful. These arguments, while intuitively compelling, are primarily philosophical. In this paper, we provide formal models that demonstrate Omohundro’s thesis, thereby putting mathematical weight behind those intuitive claims.

Russell, Bostrom and the Risk of AI


Now I suppose it is possible that once an agent reaches a sufficient level of intellectual ability it derives some universal morality from the ether and there really is nothing to worry about, but I hope you agree that this is, at the very least, not a conservative assumption. For the purposes of this article I will take the orthogonality thesis as a given. So a smarter than human artificial intelligence can have any goal.

The Electric Turing Acid Test


Parts of this essay by Andrew Smart are adapted from his book Beyond Zero And One (2015), published by OR Books. Machine intelligence is growing at an increasingly rapid pace. The leading minds on the cutting edge of AI research think that machines with human-level intelligence will likely be realized by the year 2100. Beyond this, artificial intelligences that far outstrip human intelligence would rapidly be created by the human-level AIs. This vastly superhuman AI will result from an "intelligence explosion."

Godseed: Benevolent or Malevolent? Artificial Intelligence

It is hypothesized by some thinkers that benign looking AI objectives may result in powerful AI drives that may pose an existential risk to human society. We analyze this scenario and find the underlying assumptions to be unlikely. We examine the alternative scenario of what happens when universal goals that are not human-centric are used for designing AI agents. We follow a design approach that tries to exclude malevolent motivations from AI agents, however, we see that objectives that seem benevolent may pose significant risk. We consider the following meta-rules: preserve and pervade life and culture, maximize the number of free minds, maximize intelligence, maximize wisdom, maximize energy production, behave like human, seek pleasure, accelerate evolution, survive, maximize control, and maximize capital. We also discuss various solution approaches for benevolent behavior including selfless goals, hybrid designs, Darwinism, universal constraints, semi-autonomy, and generalization of robot laws. A "prime directive" for AI may help in formulating an encompassing constraint for avoiding malicious behavior. We hypothesize that social instincts for autonomous robots may be effective such as attachment learning. We mention multiple beneficial scenarios for an advanced semi-autonomous AGI agent in the near future including space exploration, automation of industries, state functions, and cities. We conclude that a beneficial AI agent with intelligence beyond human-level is possible and has many practical use cases.