In recent months, the limitations of deep learning have been questioned by the AI community. Throughout several platforms, notably Twitter and the premier AI conference NeurIPS, AI researchers have discussed deep learning’s shortcomings. Today’s neural networks will eventually face boundaries in their abilities and usage.
One researcher stated that although machine learning holds great potential, the machines still learn in a very narrow way, requiring much more data to learn a task than a human. He also stated that AI machines still make errors and “stupid mistakes.” However, there it is important to recognize the relationship between academic research and the successful commercialization of that research, according to Forbes. Artificial intelligence does not need to surpass human capabilities to obtain commercial value.
Read More: Deep Learning Has Limits. But Its Commercial Impact Has Just Begun.