We leverage the non-fungible tokens (NFTs) setting to assess the valuation of emotional dividends (LOVE), a long-standing empirical challenge in private-value markets such as art, antiques, and collectibles. Having created and validated our proxy, we use deep learning algorithms and discover that contemporaneous price fluctuations, collection features, and ownership wealth significantly contribute to the formation of LOVE. Understanding the drivers of LOVE, we employ AI-powered algorithms to estimate the prices of NFTs. While AI models accurately predict NFT prices, the performance is decreasing in LOVE. Finally, we demonstrate that LOVE-driven trading leads to significant financial losses over the long term, suggesting that some traders trade off wealth for emotional utility. Our study provides novel economic insights into the factors behind emotional dividends and their role in the pricing of private-value assets. It also highlights the challenges AI-powered trading faces in markets with too much LOVE.