All of a sudden, it's OK to say Artificial Intelligence

For years now, companies have been hiding behind the moniker "machine learning" as a way to avoid the stigma associated (or associated in the past from the AI winters) with "artificial intelligence". Well, no more. This week Google announced Google Neural Machine Translation (GNMT) that gets close to human performance. Although Google did not self-describe it as artificial intelligence, many media outlets such as Wired and FastCompany did.

But is it really AI? Of course AI is a hopelessly ambiguous term -- referring to anything from 1950s state-space search and tree pruning for playing checkers all the way up to the tautologically impossible goal of performing a human-level task previously not done. But regardless of the definition of AI, what GNMT does is simply deep learning -- that is, neural networks; that is, machine learning. GNMT is better than previous methods only by incremental improvement in machine learning and not by a paradigm shift such as one of the four paths I previously laid out. Thus, there is no reason for a terminology change save for that the term AI has started to become socially acceptable.

Further evidence is that the inaugural O'Reilly AI Conference was held this week. Even traditionally staid IBM is paving the way for artificial intelligence -- or at least the term -- when they announced this week that the next generation of analytics will move from machine learning to artificial intelligence.

But some are more timid. There are those that try to go halfway. by using the term machine intelligence.

Finally, from the O'Reilly AI conference, if you want to jump on the AI rocket ship, one of the presentations from there tells you how.