Every hype cycle has a period in which implementation is beginning and the issues become clearer. AI is in that stage. The widespread hype over the past couple of years has been tremendous. But as the latest wave of new technologies enters into production, these issues are coming into focus.
AI is based on large volumes of data and various algorithms. The data can be used to "train" the algorithms. To do so, the data need to be not only voluminous, but clear of errors and bias. So it goes back to the quality of the data. As for the algorithms, they often start out as generalities, biased by social and economic norms that may not apply in a particular application. So the training is needed for that. Also, training is needed to enable the AI to adopt to changing circumstances. And the data must reflect those fairly.
These issues will not stop the advances of AI, nor will they slow it up very much. They just represent a normal part of the cycle - the learning cycle. They will make AI stronger in the end.
No comments:
Post a Comment