James Steinhoff

This paper sketches a prehistory of synthetic data in the development of simulation technologies. Synthetic data is connected to simulation by the technical problem of the reality gap: the gap between the synthetic data a model is trained on and the real-world data it is deployed on. The reality gap is presented as a novelty both generated and solved by synthetic data. We demonstrate that the reality gap has plagued simulation technologies since their inception in the mid-20th century. We contend that the reality gap is not something synthetic data can solve. To illustrate this, we examine three episodes in the prehistory of synthetic data. These episodes are representative of three distinct regimes of simulation: (a) the statistical regime, (b) the discrete-event regime, and (c) the visual-interactive regime. Each regime reveals a reality gap; from before the advent of digital computers to the present. Synthetic data, like simulations, require data about a given domain in order to model it. It requires the real-world data which it purports to dispense with. The reality gap is thus an epistemological issue as well as a technical one. We argue that it is also a political economic issue: it complicates existing means of producing data, adding new layers of mediation and labor. Synthetic data thus indicates the emergence of an alternative stack for the production of AI systems. This suggests that the political economy of AI must take account of the proliferation of new technical means for creating data.

This paper sketches a prehistory of synthetic data in the development of simulation technologies. Synthetic data is connected to simulation by the technical problem of the reality gap: the gap between the synthetic data a model is trained on and the real-world data it is deployed on. The reality gap is presented as a novelty both generated and solved by synthetic data. We demonstrate that the reality gap has plagued simulation technologies since their inception in the mid-20th century. We contend that the reality gap is not something synthetic data can solve. To illustrate this, we examine three episodes in the prehistory of synthetic data. These episodes are representative of three distinct regimes of simulation: (a) the statistical regime, (b) the discrete-event regime, and (c) the visual-interactive regime. Each regime reveals a reality gap; from before the advent of digital computers to the present. Synthetic data, like simulations, require data about a given domain in order to model it. It requires the real-world data which it purports to dispense with. The reality gap is thus an epistemological issue as well as a technical one. We argue that it is also a political economic issue: it complicates existing means of producing data, adding new layers of mediation and labor. Synthetic data thus indicates the emergence of an alternative stack for the production of AI systems. This suggests that the political economy of AI must take account of the proliferation of new technical means for creating data.

AI ethics is proposed, by the Big Tech companies which lead AI research and development, as the cure for diverse social problems posed by the commercialization of data-intensive technologies. It aims to reconcile capitalist AI production with ethics. However, AI ethics is itself now the subject of wide criticism; most notably, it is accused of being no more than “ethics washing” a cynical means of dissimulation for Big Tech, while it continues its business operations unchanged. This paper aims to critically assess, and go beyond the ethics washing thesis. I argue that AI ethics is indeed ethics washing, but not only that. It has a more significant economic function for Big Tech. To make this argument I draw on the theory of intellectual monopoly capital. I argue that ethics washing is better understood as a subordinated innovation network: a dispersed network of contributors beyond Big Tech’s formal employment whose research is indirectly planned by Big Tech, which also appropriates its results. These results are not intended to render AI more ethical, but rather to advance the business processes of data-intensive capital. Because the parameters of AI ethics are indirectly set in advance by Big tech, the ostensible goal that AI ethics sets for itself—to resolve the contradiction between business and ethics—is in fact insoluble. I demonstrate this via an analysis of the latest trend in AI ethics: the operationalization of ethical principles.