Which artificial intelligence (AI) systems are agents? To answer this question, I propose a multidimensional account of agency. According to this account, a system's agency profile is jointly determined by its level of goal-directedness and autonomy as well as is abilities for directly impacting the surrounding world, long-term planning and acting for reasons. Rooted in extant theories of agency, this account enables fine-grained, nuanced comparative characterizations of artificial agency. I show that this account has multiple important virtues and is more (...) informative than alternatives. More speculatively, it may help to illuminate two important emerging questions in AI ethics: 1. Can agency contribute to the moral status of non-human beings, and how? 2. When and why might AI systems exhibit power-seeking behaviour and does this pose an existential risk to humanity? (shrink)

How can one build AI systems such that they pursue the goals their designers want them to pursue? This is the alignment problem. Numerous authors have raised concerns that, as research advances and systems become more powerful over time, misalignment might lead to catastrophic outcomes, perhaps even to the extinction or permanent disempowerment of humanity. In this paper, I analyze the severity of this risk based on current instances of misalignment. More specifically, I argue that contemporary large language models and (...) game-playing agents are sometimes misaligned. These cases suggest that misalignment tends to have a variety of features: misalignment can be hard to detect, predict and remedy, it does not depend on a specific architecture or training paradigm, it tends to diminish a system’s usefulness and it is the default outcome of creating AI via machine learning. Subsequently, based on these features, I show that the risk of AI alignment magnifies with respect to more capable systems. Not only might more capable systems cause more harm when misaligned, aligning them should be expected to be more difficult than aligning current AI. (shrink)

The singularity hypothesis is a radical hypothesis about the future of artificial intelligence on which self-improving artificial agents will quickly become orders of magnitude more intelligent than the average human. Despite the ambitiousness of its claims, the singularity hypothesis has been defended at length by leading philosophers and artificial intelligence researchers. In this paper, I argue that the singularity hypothesis rests on scientifically implausible growth assumptions. I show how leading philosophical defenses of the singularity hypothesis (Chalmers 2010, Bostrom 2014) fail (...) to overcome the case for skepticism. I conclude by drawing out philosophical implications of this discussion for our understanding of consciousness, personal identity, digital minds, existential risk, and ethical longtermism. GPI Working Paper No . 19-2022. (shrink)

When asked about humanity’s future relationship with computers, Marvin Minsky famously replied “If we’re lucky, they might decide to keep us as pets”. A number of eminent authorities continue to argue that there is a real danger that “super-intelligent” machines will enslave—perhaps even destroy—humanity. One might think that it would swiftly follow that we should abandon the pursuit of AI. Instead, most of those who purport to be concerned about the existential threat posed by AI default to worrying about what (...) they call the “Friendly AI problem”. Roughly speaking this is the question of how we might ensure that the AI that will develop from the first AI that we create will remain sympathetic to humanity and continue to serve, or at least take account of, our interests. In this paper I draw on the “neo-republican” philosophy of Philip Pettit to argue that solving the Friendly AI problem would not change the fact that the advent of super-intelligent AI would be disastrous for humanity by virtue of rendering us the slaves of machines. A key insight of the republican tradition is that freedom requires equality of a certain sort, which is clearly lacking between pets and their owners. Benevolence is not enough. As long as AI has the power to interfere in humanity’s choices, and the capacity to do so without reference to our interests, then it will dominate us and thereby render us unfree. The pets of kind owners are still pets, which is not a status which humanity should embrace. If we really think that there is a risk that research on AI will lead to the emergence of a superintelligence, then we need to think again about the wisdom of researching AI at all. (shrink)

The standard argument to the conclusion that artificial intelligence (AI) constitutes an existential risk for the human species uses two premises: (1) AI may reach superintelligent levels, at which point we humans lose control (the ‘singularity claim’); (2) Any level of intelligence can go along with any goal (the ‘orthogonality thesis’). We find that the singularity claim requires a notion of ‘general intelligence’, while the orthogonality thesis requires a notion of ‘instrumental intelligence’. If this interpretation is correct, they cannot be (...) joined as premises and the argument for the existential risk of AI turns out invalid. If the interpretation is incorrect and both premises use the same notion of intelligence, then at least one of the premises is false and the orthogonality thesis remains itself orthogonal to the argument to existential risk from AI. In either case, the standard argument for existential risk from AI is not sound.—Having said that, there remains a risk of instrumental AI to cause very significant damage if designed or used badly, though this is not due to superintelligence or a singularity. (shrink)

The last 5 years have seen a series of remarkable achievements in deep-neural-network-based artificial intelligence research, and some modellers have argued that their performance compares favourably to human cognition. Critics, however, have argued that processing in deep neural networks is unlike human cognition for four reasons: they are (i) data-hungry, (ii) brittle, and (iii) inscrutable black boxes that merely (iv) reward-hack rather than learn real solutions to problems. This article rebuts these criticisms by exposing comparative bias within them, in the (...) process extracting some more general lessons that may also be useful for future debates. (shrink)

This paper looks at philosophical questions that arise in the context of AI alignment. It defends three propositions. First, normative and technical aspects of the AI alignment problem are interrelated, creating space for productive engagement between people working in both domains. Second, it is important to be clear about the goal of alignment. There are significant differences between AI that aligns with instructions, intentions, revealed preferences, ideal preferences, interests and values. A principle-based approach to AI alignment, which combines these elements (...) in a systematic way, has considerable advantages in this context. Third, the central challenge for theorists is not to identify ‘true’ moral principles for AI; rather, it is to identify fair principles for alignment that receive reflective endorsement despite widespread variation in people’s moral beliefs. The final part of the paper explores three ways in which fair principles for AI alignment could potentially be identified. (shrink)

Recent progress in artificial intelligence has renewed interest in building systems that learn and think like people. Many advances have come from using deep neural networks trained end-to-end in tasks such as object recognition, video games, and board games, achieving performance that equals or even beats that of humans in some respects. Despite their biological inspiration and performance achievements, these systems differ from human intelligence in crucial ways. We review progress in cognitive science suggesting that truly human-like learning and thinking (...) machines will have to reach beyond current engineering trends in both what they learn and how they learn it. Specifically, we argue that these machines should build causal models of the world that support explanation and understanding, rather than merely solving pattern recognition problems; ground learning in intuitive theories of physics and psychology to support and enrich the knowledge that is learned; and harness compositionality and learning-to-learn to rapidly acquire and generalize knowledge to new tasks and situations. We suggest concrete challenges and promising routes toward these goals that can combine the strengths of recent neural network advances with more structured cognitive models. (shrink)

This paper discusses the relation between intelligence and motivation in artificial agents, developing and briefly arguing for two theses. The first, the orthogonality thesis, holds (with some caveats) that intelligence and final goals (purposes) are orthogonal axes along which possible artificial intellects can freely vary—more or less any level of intelligence could be combined with more or less any final goal. The second, the instrumental convergence thesis, holds that as long as they possess a sufficient level of intelligence, agents having (...) any of a wide range of final goals will pursue similar intermediary goals because they have instrumental reasons to do so. In combination, the two theses help us understand the possible range of behavior of superintelligent agents, and they point to some potential dangers in building such an agent. (shrink)

The human brain has some capabilities that the brains of other animals lack. It is to these distinctive capabilities that our species owes its dominant position. Other animals have stronger muscles or sharper claws, but we have cleverer brains. If machine brains one day come to surpass human brains in general intelligence, then this new superintelligence could become very powerful. As the fate of the gorillas now depends more on us humans than on the gorillas themselves, so the fate of (...) our species then would come to depend on the actions of the machine superintelligence. But we have one advantage: we get to make the first move. Will it be possible to construct a seed AI or otherwise to engineer initial conditions so as to make an intelligence explosion survivable? How could one achieve a controlled detonation? To get closer to an answer to this question, we must make our way through a fascinating landscape of topics and considerations. Read the book and learn about oracles, genies, singletons; about boxing methods, tripwires, and mind crime; about humanity's cosmic endowment and differential technological development; indirect normativity, instrumental convergence, whole brain emulation and technology couplings; Malthusian economics and dystopian evolution; artificial intelligence, and biological cognitive enhancement, and collective intelligence. (shrink)

What happens when machines become more intelligent than humans? One view is that this event will be followed by an explosion to ever-greater levels of intelligence, as each generation of machines creates more intelligent machines in turn. This intelligence explosion is now often known as the “singularity”. The basic argument here was set out by the statistician I.J. Good in his 1965 article “Speculations Concerning the First Ultraintelligent Machine”: Let an ultraintelligent machine be defined as a machine that can far (...) surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an “intelligence explosion”, and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make. The key idea is that a machine that is more intelligent than humans will be better than humans at designing machines. So it will be capable of designing a machine more intelligent than the most intelligent machine that humans can design. So if it is itself designed by humans, it will be capable of designing a machine more intelligent than itself. By similar reasoning, this next machine will also be capable of designing a machine more intelligent than itself. If every machine in turn does what it is capable of, we should expect a sequence of ever more intelligent machines. This intelligence explosion is sometimes combined with another idea, which we might call the “speed explosion”. The argument for a speed explosion starts from the familiar observation that computer processing speed doubles at regular intervals. Suppose that speed doubles every two years and will do so indefinitely. Now suppose that we have human-level artificial intelligence 1 designing new processors. Then faster processing will lead to faster designers and an ever-faster design cycle, leading to a limit point soon afterwards. The argument for a speed explosion was set out by the artificial intelligence researcher Ray Solomonoff in his 1985 article “The Time Scale of Artificial Intelligence”.1 Eliezer Yudkowsky gives a succinct version of the argument in his 1996 article “Staring at the Singularity”: “Computing speed doubles every two subjective years of work.. (shrink)