We explore whether quantum field theory can be understood as the statistical mechanics of a time-reversal-invariant stochastic generalization of Hamiltonian dynamics. The motivation for this project, started with this paper, is to assign sharp values to all observables and thereby avoid the quantum measurement problem. In classical mechanics, motion is deterministic and corresponds to an evolution of the phase space probability density according to Liouville's equation that is governed by first derivatives of the Hamiltonian in phase space. We derive a generalization of the Liouville equation with natural constraints – namely, reduction to classical Hamiltonian dynamics as the stochasticity parameter ħ↦0, Fokker-Planck form for the probability density evolution, local Hamiltonian dependence, time-reversal invariance, energy conservation, and minimality – which turns out to be a Fokker-Planck equation with a generalized diffusion matrix that is symmetric, traceless, and constructed from the Hessian of the Hamiltonian. We then show that the Schrödinger equation in the coherent-state phase-space formulation of certain bosonic QFTs has precisely this form, with the Husimi function playing the role of the phase space probability density. The question to what extent this equation can be interpreted in terms of objective stochastic field theories is discussed in a companion paper.

The Kochen-Specker theorem shows that it is impossible to assign sharp values to all dynamical variables in quantum mechanics in such a way that the algebraic relations among the values of dynamical variables whose self-adjoint operators commute are the same as those among the operators themselves. We point out that, for quantum theories obtained by quantizing some classical theory, this condition-Kochen-Specker non-contextuality-is implausible from the start because quantization usually changes algebraic relations. We explain why this is so, using the formalism of deformation quantization and its conception of star products, and we illustrate the relevance of this point using various examples of dynamical variables quantized via Weyl quantization and coherent state quantization. Our observations suggest that the relevance of the Kochen-Specker theorem to the question of whether one can assign sharp values to all dynamical variables is rather limited.

This paper proposes an approach to interpreting quantum expectation values that may help address the quantum measurement problem. Quantum expectation values are usually calculated via Hilbert space inner products and, thereby, differently from expectation values in classical mechanics, which are weighted phase-space integrals. It is shown that, by using Anti-Wick quantization to associate dynamical variables with self-adjoint linear operators, quantum expectation values can be interpreted as genuine weighted averages over phase space, paralleling their classical counterparts. This interpretation arises naturally in the Segal-Bargmann space, where creation and annihilation operators act as simple multiplication and differentiation operators. In this setting, the Husimi Q-function - the coherent-state representation of the quantum state - can be seen as a true probability density in phase space. Unlike Bohmian mechanics, the present approach retains the standard correspondence between dynamical variables and self-adjoint operators while paving the way for a classical-like probabilistic interpretation of quantum statistics.

In a companion paper we derived a unique time-reversal-invariant stochastic generalization of the Liouville equation and showed that it coincides with the evolution equation for the Husimi Q-function in a broad class of bosonic quantum field theories. Here we investigate the prospects for interpreting that evolution equation in terms of underlying stochastic trajectories. Drawing on Drummond's time-symmetric stochastic action formalism, we show that the traceless diffusion Fokker-Planck equation defines a natural measure over stochastic trajectories conditional on mixed-time boundary conditions. However, we identify a significant gap: it has not been established that every Q-function can be represented as a weighted average of these conditional probabilities over boundary values. The trajectory interpretation holds for ensembles with fixed boundary conditions but does not straightforwardly extend to arbitrary quantum states. Despite this limitation, we show that Drummond's trajectory dynamics are fundamentally non-Markovian – a natural consequence of combining stochasticity with time-reversal invariance. This non-Markovianity places the dynamics outside the scope of the ontological models framework and thereby explains why the major no-go theorems for hidden-variable theories do not rule out the approach. These results clarify both the achievements and the remaining challenges in the project of understanding quantum field theory as the statistical mechanics of time-symmetric stochastic processes.

Agents are said to be “clueless” if they are unable to predict some ethically important consequences of their actions. Some philosophers have argued that such “cluelessness'' is widespread and creates problems for certain approaches to ethics. According to Hilary Greaves, a particularly problematic type of cluelessness, namely, “complex” cluelessness, affects attempts to do good as effectively as possible, as suggested by proponents of “Effective Altruism,” because we are typically clueless about the long-term consequences of such interventions. As a reaction, she suggests focusing on interventions that are long-term oriented from the start. This paper argues for three claims: first, that David Lewis’ distinction between sensitive and insensitive causation can help us better understand the differences between genuinely “complex” and more harmless “simple” cluelessness; second, that Greaves’ worry about complex cluelessness can be mitigated for attempts to do near-term good; and, third, that Greaves’ recommendation to focus on long term-oriented interventions in response to complex cluelessness is not promising as a strategy specifically for avoiding complex cluelessness. There are systematic reasons why the actual effects of serious attempts to beneficially shape the long-term future are inherently difficult to predict and why, hence, such attempts are prone to backfiring.

In response to the worry that autonomous generally intelligent artificial agents may at some point take over control of human affairs a common suggestion is that we should “solve the alignment problem” for such agents. We show that current discourse around this suggestion often uses a particular framing of artificial intelligence (AI) alignment as binary, a natural kind, mainly a technical‐scientific problem, realistically achievable, or clearly operationalizable. Each of these assumptions may not actually be true. We further argue that this “Manhattan project framing” of AI alignment may bias societal discourse and decision‐making towards faster AI development and deployment than is responsible.

Bell's theorem states that no model that respects Local Causality and Statistical Independence can account for the correlations predicted by quantum mechanics via entangled states. This paper proposes a new approach, using backward-in-time conditional probabilities, which relaxes conventional assumptions of temporal ordering while preserving Statistical Independence as a "fine-tuning condition. It is shown how such models can account for EPR/Bell correlations and, analogously, the GHZ predictions while nevertheless forbidding superluminal signalling.

This article outlines a novel interpretation of quantum theory: the Q-based interpretation. The core idea underlying this interpretation, recently suggested for quantum field theories by Drummond and Reid (2020), is to interpret the phase space function Q---a transform of the better known Wigner function---as a proper probability distribution, roughly analogous to the probability distribution \rho in classical statistical mechanics. Here I motivate the Q-based interpretation, investigate whether it is empirically adequate, and outline some of its key conceptual features. I argue that the Q-based interpretation is attractive in that it promises having no measurement problem, is conceptually parsimonious and has the potential to apply elegantly to relativistic and field-theoretic contexts.

Some philosophers and machine learning experts have speculated that superintelligent Artificial Intelligences (AIs), if and when they arrive on the scene, will wrestle away power from humans, with potentially catastrophic consequences. Dan Hendrycks has recently buttressed such worries by arguing that AI systems will undergo evolution by natural selection, which will endow them with instinctive drives for self-preservation, dominance and resource accumulation that are typical of evolved creatures. In this paper, we argue that this argument is not compelling as it stands. Evolutionary processes, as we point out, can be more or less Darwinian along a number of dimensions. Making use of Peter Godfrey-Smith’s framework of Darwinian spaces, we argue that the more evolution is top-down, directed and driven by intelligent agency, the less paradigmatically Darwinian it becomes. We then apply the concept of “domestication” to AI evolution, which, although theoretically satisfying the minimal definition of natural selection, is channeled through the minds of fore-sighted and intelligent agents, based on selection criteria desirable to them (which could be traits like docility, obedience and non-aggression). In the presence of such intelligent planning, it is not clear that selection of AIs, even selection in a competitive and ruthless market environment, will end up favoring “selfish” traits. In the end, however, we do agree with Hendrycks’ conditionally: If superintelligent AIs end up “going feral” and competing in a truly Darwinian fashion, reproducing autonomously and without human supervision, this could pose a grave danger to human societies.

A transition to a world with artificial general intelligence (AGI) may occur within the next few decades. This transition may give rise to catastrophic risks from misaligned AGI, which have received a significant amount of attention, deservedly. Here I argue that AGI systems that are intent-aligned —they always try to do what their operators want them to do—would also create catastrophic risks, mainly due to the power that they concentrate on their operators. With time, that power would almost certainly be catastrophically exploited, potentially resulting in human extinction or permanent dystopia. I suggest that liberal democracies, if they decide to allow the development of AGI, may react to this threat by letting AGI take shape as an intergenerational social project , resulting in an arrangement where AGI is not intent-aligned but symbiotic with humans. I provide some tentative ideas on what the resulting arrangement may look like and consider what speaks for and what against aiming for intent-aligned AGI as an intermediate step.

If and when artificial intelligence systems become superhuman in more aspects of analytic reasoning, this will inevitably have a strong impact on the social organisation of science, including academic writing, reviewing, and publishing. We consider how norms of academic publishing should be adjusted as this happens. To do so, we propose four intuitively plausible desiderata that norms of academic publishing should fulfil in the age of increasingly advanced artificial intelligence (AI) and argue that there are no “quick fixes” to current norms that fulfil these desiderata. To indicate the scale of change needed to prepare academic publishing for the era of increasingly advanced AI, we tentatively sketch a more promising novel system of norms. Our proposal centres around the idea that AI systems should “sign off’’ on statements that outline the human and AI contributions to academic research. We discuss possible challenges for this proposal and highlight the type of technological and regulatory infrastructure that would be needed to enable it.

Policy instruments promoting sustainability, such as investment taxonomies, are playing an increasing role in guiding the allocation of financial resources internationally. But can policy instruments define sustainability in ways that are both operational (i.e. assessable via replicable procedures) and which specify practices that can reliably be expected to enhance future generations' welfare? This paper analyses candidate definitions of sustainability and identifies a dilemma: while various definitions identify a 'capital' variable whose value can indeed be determined empirically; we have no reason to assume that preservation of any specific capital variable will maximise expected future welfare. By contrast, sustainability can be defined 'dynamically' in terms of activities that will, on expectation, lead to future developmental trajectories with high welfare. But, as we show through discussion of concrete examples, 'dynamic sustainability' cannot readily be operationalised. We conclude that what qualifies as 'sustainable' will remain a subject of political dispute and that authoritative comprehensive assessments of 'sustainability' will remain chimeric. We suggest that selecting a narrow class of specific measures, such as of life-cycle greenhouse gas emissions, might lead to more effective and less contentious approaches to resource allocation.

Recent astrophysical findings suggest that the era during which the Universe is habitable has just begun. This raises the question whether the entire Universe may at some point in the future be filled with intelligent life. Hanson et al . (2021, The Astrophysical Journal 922 , 182) argued that we can be confident that the Universe will, by cosmic standards, soon be dominated by imperialist civilizations which expand rapidly, persist long and make drastic changes to the volumes they control. The main motivation for this ‘grabby civilizations’ hypothesis is that it supposedly provides a good explanation of why we are so early in cosmic history. In this paper, we criticize this motivation and suggest that it fails, for reasons analogous to why the notorious Doomsday argument fails. In the last part of the paper we broaden our discussion and argue that it may be rational to assign a rather low prior probability to the grabby civilizations hypothesis. For instance, if there are any civilizations that expand rapidly and indefinitely, they may well not make any drastic changes to the volumes they inhabit, potentially for strategic reasons. Hence, we call for epistemic caution and humility regarding the question of the long-term evolution of intelligence in the Universe.

Gauge symmetries play a central role, both in the mathematical foundations as well as the conceptual construction of modern (particle) physics theories. However, it is yet unclear whether they form a necessary component of theories, or whether they can be eliminated. It is also unclear whether they are merely an auxiliary tool to simplify (and possibly localize) calculations or whether they contain independent information. Therefore their status, both in physics and philosophy of physics, remains to be fully clarified. This Element reviews the current state of affairs on both the philosophy and the physics side. In particular, it focuses on the circumstances in which the restriction of gauge theories to gauge invariant information on an observable level is warranted, using the Brout-Englert-Higgs theory as an example of particular current importance. Finally, the authors determine a set of yet to be answered questions to clarify the status of gauge symmetries.

Global energy justice remains far out of reach. If the goal of energy justice is the universal, equitable, and democratic provision of safe, affordable, and sustainable energy services, the international community currently lacks the physical, ideational, or governance infrastructure necessary for its realization. Instead, access to energy remains radically unequal, continuing greenhouse gas emissions are creating intergenerational sabotage, and fossil fuel revenues routinely corrupt democratic politics. In addition to distributive injustice, global energy governance also creates dilemmas of procedure and recognition that are our focus. Here, we first identify inherent tensions between local democratic sovereignty and global energy justice and then argue that existing energy governance infrastructures often amplify powerful actors' leverage over the energy choices and strategies of less powerful communities. We conclude by discussing the design of a governance infrastructure that could promote climate mitigation and energy access goals without exploiting international inequalities in ways that risk undermining justice.

In recent years, there has been an intense public debate about whether and, if so, to what extent investments in nuclear energy should be part of strategies to mitigate climate change. Here, we address this question from an ethical perspective, evaluating different strategies of energy system development in terms of three ethical criteria, which will differentially appeal to proponents of different normative ethical frameworks. Starting from a standard analysis of climate change as arising from an intergenerational collective action problem, we evaluate whether contributions from nuclear energy will, on expectation, increase the likelihood of successfully phasing out fossil fuels in time to avert dangerous global warming. For many socio-economic and geographic contexts, our review of the energy system modeling literature suggests the answer to this question is “yes.” We conclude that, from the point of view of climate change mitigation, investments in nuclear energy as part of a broader energy portfolio will be ethically required to minimize the risks of decarbonization failure, and thus the tail risks of catastrophic global warming. Finally, using a sensitivity analysis, we consider which other aspects of nuclear energy deployment, apart from climate change, have the potential to overturn the ultimate ethical verdict on investments in nuclear energy. Out of several potential considerations (e.g., nuclear waste, accidents, safety), we suggest that its potential interplay — whether beneficial or adverse — with the proliferation of nuclear weapons is the most plausible candidate.

The most-discussed objection against this argument is that it commits the inverse gambler's fallacy, originally identified by Ian Hacking. This fallacy consists in inferring from an event with a remarkable outcome that there have likely been many more events of the same type in the past, most with less remarkable outcomes. I discuss several suggested analogs to the problem of the fine-tuned parameters. Ultimately, as I argue, established standards of rationality may just not allow one to decide whether the standard fine-tuning argument for the multiverse commits the inverse gambler’s fallacy or not. Some of the considerations in this chapter, as explained along the way, are relevant to the debate about the Fermi paradox.