Alfred P. Sloan Foundation

Recent advancements in generative AI (GenAI) have ignited debates about its impact on the future of work, industry, and society – particularly because, unlike previous technologies, GenAI can generate creative content. This chapter explores the implications of GenAI for creative work across industries by distinguishing between two types of creativity: scientific and artistic. Drawing on interdisciplinary literature, we argue that while GenAI can complement both, it is more likely to substitute humans in tasks involving artistic rather than scientific creativity. This is because scientific creativity relies more heavily on causal inference – a capability that current GenAI systems lack. Building on this foundation, we examine how GenAI is reshaping industries and propose two factors that moderate its impact: the importance of basic research to an industry and its exposure to AI. The chapter offers strategic insights for practitioners and policymakers to better navigate GenAI’s evolving role in fostering creative work across industries.

The final version of this paper will appear as Chapter 6 in the forthcoming MIT Energy Initiative study, The Future of Storage. Chapters referred to in this paper will be included in that study when it is published. In order to illuminate the role of energy storage in future decarbonized electric power systems, we construct detailed models, calibrated to mid-century, of optimal assets and hourly operation of power systems under a range of assumptions about generation and storage technologies’ availability and cost. We model three US regions: The Northeast, the Southeast, and Texas. These regions differ in many dimensions, notably in the quality of their variable renewable energy (VRE, wind and solar) resource and load profiles. We find that nearly complete decarbonization of all three systems using only VRE generation and (very little) natural gas, along with Lithium-ion storage, can be achieved without reduced reliability or very large increases in system average electricity cost. The incremental cost of going to complete decarbonization of the electric power system without any offsets from other sectors is very high, however, comparable or higher than estimated costs of negative emissions technologies. If technologies more suitable for long-duration storage are available, they optimally substitute for dispatchable natural gas capacity and, under plausible assumptions, produce only moderate reductions in system average electricity cost. Substantial industrial demand for hydrogen would make its use for storage in the electric power system more attractive. In decarbonized power systems, the distribution of the hourly marginal value of energy (MVE), which corresponds roughly to the wholesale spot price, will be drastically different from the distributions of spot prices in current systems: there will be more hours of high MVEs and many more hours of very low MVEs. In order to encourage efficient economy-wide decarbonization, wholesale markets and retail rate structures will need to be significantly modified. In addition, research in the design, operation, and regulation of decarbonized systems should be a high priority.

Entrepreneurs, managers, and scientists participate in competitive selection processes to obtain resources. The project they propose is a crucial aspect of their success. In this paper, we focus on the selection of scientists applying for academic funding by submitting a research proposal. We argue that two core dimensions of the research proposal affect the probability of funding success: its coherence with the applicant’s previous work, and its alignment with subjects of general interest for the scientific community. Employing a neural network algorithm, we analyse the text of 2,494 research proposals for a prestigious fellowship awarded to promising early-stage North American researchers. We find field-specific heterogeneity in the committees’ evaluations. In life sciences and chemistry, evaluators value the research proposal’s coherence positively with the scientist’s recent work and the proposals’ alignment with the current subject of general interest for the scientific community. Conversely, in physics, evaluators give more weight to bibliometric indicators and less to the proposal coherence and alignment. Our results can be extended beyond the academic context to managerial implications in cases such as entrepreneurs and managers submitting project proposals to investors.

Central to all innovation processes is the selection of candidates with the highest potential to succeed. The selection procedure is usually based on an analysis of the signal given by the candidate's past endeavors as well as the projects proposed. In this paper, we focus on the selection of scientists applying for academic funding of their research proposals. We argue that two core dimensions of a scientist's application affect the probability of funding success: the coherence of the proposal submitted with the applicant's previous work, and the alignment of the proposal with articles published in top generalist science journals. Employing a neural network algorithm, we analyze the text of 2,494 research proposals of applications for the Sloan Research Fellowship, a prestigious fellowship awarded to promising early-stage North American researchers. We find that, on average, scientists are rewarded when they pursue research aligned with the current general interests of the scientific community. We also find field-specific heterogeneity in the type of scientific features rewarded by an evaluation committee. In life sciences and chemistry, evaluators value coherent research agendas, particularly coherence between a research proposal and the scientist's recent work, and place little importance on bibliometric achievements. Conversely, in physics, examiners give more weight to bibliometric indicators and less to applicants' research coherence. Our results show that the degree of coherence in a scientist's research agenda acts as a strong signal in the screening process. Managerial implications can also be derived for entrepreneurs and managers submitting project proposals to investors.