How does limited attention affect stock prices in today’s computer-driven financial markets? We study this issue by re-examining the effects of limited attention using a dataset that separately identifies trades made by high-frequency traders (HFTs, or computers) versus those made by non-high-frequency traders (human decision-makers). We employ a set of six attention proxies to identify earnings announcements with low investor attention: announcements made on Fridays and on days with multiple earnings announcements, and announcements with slow analyst forecast adjustments, high news distraction, low EDGAR download volume, and low Google search volume. Across multiple attention proxies, we find that HFT trading improves the responsiveness of prices by increasing the short-horizon price response and reducing the long-term price drift following earnings surprises, diminishing the inefficiencies previously observed around low-attention announcements by 69% to 100%. We find that the price efficiency improvements are more closely tied to HFT liquidity demand than supply, suggesting that HFTs improve efficiency by processing and trading on the information in low-attention announcements.