How to kickstart an AI venture without proprietary data: AI start-ups have a chicken and egg problem — here is how to solve it

Click the button below to download the full text of the article.

Abstract: Even when entrepreneurs have innovative ideas for applying AI to real-world problems, they can encounter a unique challenge to kickstarting their AI ventures. Today’s AI systems need to be trained on large datasets, which poses a chicken-and-egg problem for entrepreneurs. Established companies with a sizable customer base already have a stream of data from which they can train AI systems, build new products and enhance existing ones, generate additional data, and rinse and repeat. Entrepreneurs have not yet built their company, so they do not have data, which means they cannot create an AI product as easily; however, this challenge can be navigated with a strategic approach. This paper presents five strategies that can help entrepreneurs access the data they need to break into the AI space, as well as examples of how these strategies have been used by other companies, particularly in their early stages. Specifically, the paper discusses how entrepreneurs can: 1) start by offering a service that has value without AI and that generates data; 2) partner with a non-tech company that has a proprietary dataset; 3) crowdsource the (labelled) data they need; 4) make use of public data; and 5) rethink the need for data entirely and instead use expert systems or reinforcement learning to kickstart their AI ventures.

Keywords: artificial intelligence (AI), machine learning (ML), data, entrepreneurship, innovation, AI start-ups, technology

Kartik Hosanagar is a professor at the Wharton School of the University of Pennsylvania. He is an entrepreneur and founder of Jumpcut Media and Yodle. Kartik is the author of A Human’s Guide to Machine Intelligence, and faculty director of Wharton’s AI for Business Initiative.

Monisha Gulabani is a Research Assistant at Wharton AI for Business.