Transformative research focus considered harmful

Researchers are often encouraged to pursue nothing short of revolutionary advances, and those who work in artificial intelligence no exception. However, an exclusive focus on breakthroughs is counterproductive science. As explained by Kuhn almost 50 years ago, dramatic usually rely a foundation less which uncover anomalies make marginal improvements current efforts. Progress relies essential tension between convergent divergent thinking, each being complementary aspects the same process. We argue that overemphasis on, rewarding of, thinking contemporary AI—whether form rejecting funding for nontransformative research, or peer-review criteria papers lack novelty—is machine learning may even be fundamentally harmful progress field. To reckon with this problem, we recommend increasing iterative improvement theories, better guidance reviewers, more transparency public funding.