Too many AI researchers think real-world problems are not relevant

• “it is an application and the significance seems limited for the machine-learning community.”” ⤴️
• “If the community feels that aiming to solve high-impact real-world problems with machine learning is of limited significance, then what are we trying to achieve?” ⤴️
• “this hyperfocus on novel methods leads to a scourge of papers that report marginal or incremental improvements” ⤴️
• “Kiri Wagstaff: “Much of current machine learning research has lost its connection to problems of import to the larger world of science and society.”” ⤴️
• “many papers that describe new applications present both novel concepts and high-impact results. But even a hint of the word “application” seems to spoil the paper for reviewers.” ⤴️
• “More than half of the images in ImageNet (pdf) come from the US and Great Britain” ⤴️
• “adapting machine-learning tools to specific real-world problems takes significant algorithmic and engineering work.” ⤴️
• “most studies applying deep learning to echocardiogram analysis try to surpass a physician’s ability to predict disease. But predicting normal heart function (pdf) would actually save cardiologists more time by identifying patients who do not need their expertise.” ⤴️
• “Many studies applying machine learning to viticulture aim to optimize grape yields (pdf), but winemakers “want the right levels of sugar and acid, not just lots of big watery berries,”” ⤴️
• “the field’s benchmark data sets are woefully out of touch with reality.” ⤴️
• “New machine-learning models are measured against large, curated data sets that lack noise and have well-defined, explicitly labeled categories (cat, dog, bird). Deep learning does well for these problems because it assumes a largely stable world (pdf).” ⤴️
• “While researchers try to outdo one another on contrived benchmarks, one in every nine people in the world is starving. Earth is warming and sea level is rising at an alarming rate.” ⤴️