Is your company’s AI getting smarter every day? It should be
Imagine working with someone who was completely unable to learn, to gain any experience whatsoever. There’s no need to picture someone completely stupid: just think of someone who joins the company, starts to deal with a lot of customers, experiences successes and failures, attends meetings, makes decisions . . . but one year later, has the same skills and capabilities as on day one. Would you be…
Companies today are deploying AI systems that do not improve over time, functioning similarly to individuals who retain their skills and capabilities without learning from experience. While AI systems can accumulate memory and retrieve previous interactions, they do not necessarily become more adept at making decisions based on past actions.
Satya Nadella of Microsoft has emphasized the distinction between a general model that can be replaced and the "company veteran" expertise that should accumulate around it. To develop truly intelligent systems, corporate AI must learn from the consequences of its actions, turning daily interactions into valuable lessons. Memory provides context, while learning drives improvement in future actions.
A research field called continual learning aims to enable AI systems to continuously learn from changing streams of experience while retaining previously acquired knowledge. The key to unlocking the true potential of corporate AI lies in building a learning loop where both human and AI expertise can compound. Unlike traditional software that merely processes large amounts of data, an intelligent system should become better prepared for subsequent interactions, recognizing patterns and refining its performance over time.
Successful recommendation systems utilized by companies like YouTube, Spotify, and Amazon have long harnessed streams of user behavior and feedback to refine future recommendations, even without AI. However, corporate AI should not be limited to simply storing data or memory. Executives need to ask themselves profound questions about what success looks like for their AI systems and how they will learn from both successes and failures.
Ultimately, the real value of corporate AI lies in its ability to evolve and improve daily, not just sit idle with a static model.
Written by urgent.news from Fast Company's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
