How long can Google keep kicking the AI can down the road?

When OpenAI was set up in 2015, Google had already been thinking for some years about developing machine learning in an environment with increasingly powerful and ubiquitous computing resources, had already bought Demis Hassabis’ Deep Mind, and was training all its staff in these technologies.

However, that advantage, following Clayton Christensen’s fantastic book “The Innovator’s Dilemma” to the letter, did not do it much good: as the leading search engine it resisted implementing AI into its products, despite claiming it was “rethinking them all” so as to do so, and that led to what we now know happened: a much smaller, younger company like OpenAI and another more veteran but largely overlooked company, Microsoft, overtook Google on the inside lane.

The reasons for Google’s hesitancy? The first and most obvious was a nascent technology that was not yet fully under control and that tended to create information from barely meaningful correlations, that meant Google was at great risk of ruining its credibility than its competitors were. What happens if the search engine you use for just about everything, overnight, starts “hallucinating” and telling you to put glue on your pizza, run around with scissors in your hand while laughing, or change the oil in your speakers? Well, chances are that instead of just laughing and…