Consider two deep-rooted notions in Silicon Valley: software is eating the world and more data beats better algorithms. Both have merit, but they don’t give the full picture. Deep learning (DL) works well when you have large, carefully labeled datasets to use for training, but not every organization has data assets like those of Google, Amazon, and Apple. To leverage machine learning while cultivating those assets needed for DL—in other words, table stakes for AI—one excellent approach is active learning, a variant of machine learning that incorporates human-in-the-loop computing. Active learning blends human intelligence and judgement with algorithms and data for a best-of-breed approach to AI in the enterprise. Many enterprise organizations are at a stage appropriate for this kind of work. One interesting benefit is that active learning focuses input from human experts. On the one hand, it leverages the human intelligence that’s already in the system: customer support, sales teams, professional services, etc. To paraphrase Amazon’s Werner Vogels, “There’s no compression algorithm for experience.” On the other hand, active learning provides systematic ways to explore and exploit the uncertainty within your data, identifying likely opportunities for profit. Paco Nathan offers an introduction to active learning and human in the loop for AI, making the business case for how and when to use it—plus many pointers to material for deep-dive study and lots of Q&A. You’ll leave with new ideas that you can immediately use to get your team talking, thinking, and taking action.