arXiv subjects

Abhishek Bhatia

Publications and source records attributed to Abhishek Bhatia.

2 recordsLinked to original sources

arXiv2025

AI Coding Tools and Digital Entrepreneurship: The Role of Software Expertise

While digital technologies expand entrepreneurial access by providing technical resources, it is less known whether they enable ventures to create durable value and whether they can substitute for technical expertise accumulated through software work experience. This paper studies how digital ventures respond to the wide diffusion of AI coding tools, and how these responses are shaped by founders' software expertise. Measuring product-category exposure to AI coding using pre-LLM product descriptions and linking venture launch, traffic, and financing data, we show that exposure to AI coding increases first-time venture launches after 2022Q4, while entrants are less likely to survive but raise more financing conditional on one-year survival. Crucially, founders with software work experience drive a larger fraction of new launches, ameliorate the decline in survival when product development is partially rather than fully automated, and explain all of the increase in venture financing.

econ.GN↗

arXiv2021

SemIE: Semantically-aware Image Extrapolation

We propose a semantically-aware novel paradigm to perform image extrapolation that enables the addition of new object instances. All previous methods are limited in their capability of extrapolation to merely extending the already existing objects in the image. However, our proposed approach focuses not only on (i) extending the already present objects but also on (ii) adding new objects in the extended region based on the context. To this end, for a given image, we first obtain an object segmentation map using a state-of-the-art semantic segmentation method. The, thus, obtained segmentation map is fed into a network to compute the extrapolated semantic segmentation and the corresponding panoptic segmentation maps. The input image and the obtained segmentation maps are further utilized to generate the final extrapolated image. We conduct experiments on Cityscapes and ADE20K-bedroom datasets and show that our method outperforms all baselines in terms of FID, and similarity in object co-occurrence statistics.

cs.CV↗