Machine Learning and Intellectual Debt

[zakiyatasnim]

In the past, intellectual debt was limited to a few fields that were accustomed to trial-and-error discovery, such as medicine. But that could change as new techniques in artificial intelligence, particularly machine learning, increase our collective intellectual credit line. Machine learning systems work by identifying patterns in oceans of data. Using these patterns, they dare to answer complex, open-ended questions. Feed a neural network labeled images of cats and other non-cat objects, and it will learn to distinguish cats from everything else; give it access to medical records, and it will attempt to predict the likelihood of a new patient dying. Yet uae number data most machine learning systems do not uncover causal mechanisms. They are statistically correlative methods. They cannot explain why they think some patients are more likely to die, because they do not “think” in any everyday sense of the word – they only answer. Once we begin to integrate their ideas into our lives, we will collectively begin to accumulate more and more intellectual debt.

Advances in theory-free pharmaceuticals show us that in some cases, intellectual debt can be vital. Millions of lives have been saved by interventions we fundamentally do not understand. Few people would refuse to take a life-saving drug—or aspirin, for that matter—simply because no one knows how it works. But accumulating intellectual debt has its downsides.

As drugs with unknown mechanisms of action proliferate, the number of tests needed to detect adverse interactions must increase exponentially (if the principles by which drugs work were understood, adverse interactions could be predicted in advance). In practice, therefore, interactions are discovered after new drugs are introduced, fueling a cycle in which drugs are introduced and then excommunicated, with class action lawsuits in the process. In any given case, the accumulation of intellectual debt associated with a new drug may be reasonable. But intellectual debts do not exist in isolation. Theory-less answers, found and deployed in different fields, can complicate each other in unpredictable ways.