When I formulated the paradigm of Jolting Technologies, I had very few data points to start from.
I looked at the increasing acceleration in the power of AI applications. In particular I criticized Stanford University’s approach which used a linear interpolation on a logarithmic chart to conclude that AI had “two eras”. My prediction was that rather than conveniently staying on the line that they drew two years ago, AI applications would evolve with an increasing acceleration.
In its recent developers’ conference, NVIDIA proved me right, quoting now a doubling rate of two months for AI power, rather than the four months reported by Stanford two years ago.
So, if we want to go out on a limb, we could talk about the law of Jolting AI, of a two year doubling in the rate of acceleration. Based on that we can predict that around 2022 the doubling rate of AI power will be about one month.
Still very few data points, and vague, rather than precise predictions. However what matters is the principle of jolting: an increasing rate of acceleration characterizing artificial intelligence applications.
What will be the consequences of this?
NVIDIA Proves That AI Is Jolting
Jolting Technologies are those characterized by an increasing rate of acceleration.
When I formulated the paradigm of Jolting Technologies, I had very few data points to start from.
I looked at the increasing acceleration in the power of AI applications. In particular I criticized Stanford University’s approach which used a linear interpolation on a logarithmic chart to conclude that AI had “two eras”. My prediction was that rather than conveniently staying on the line that they drew two years ago, AI applications would evolve with an increasing acceleration.
In its recent developers’ conference, NVIDIA proved me right, quoting now a doubling rate of two months for AI power, rather than the four months reported by Stanford two years ago.
So, if we want to go out on a limb, we could talk about the law of Jolting AI, of a two year doubling in the rate of acceleration. Based on that we can predict that around 2022 the doubling rate of AI power will be about one month.
Still very few data points, and vague, rather than precise predictions. However what matters is the principle of jolting: an increasing rate of acceleration characterizing artificial intelligence applications.
What will be the consequences of this?
When I formulated the paradigm of Jolting Technologies, I had very few data points to start from.
I looked at the increasing acceleration in the power of AI applications. In particular I criticized Stanford University’s approach which used a linear interpolation on a logarithmic chart to conclude that AI had “two eras”. My prediction was that rather than conveniently staying on the line that they drew two years ago, AI applications would evolve with an increasing acceleration.
In its recent developers’ conference, NVIDIA proved me right, quoting now a doubling rate of two months for AI power, rather than the four months reported by Stanford two years ago.
So, if we want to go out on a limb, we could talk about the law of Jolting AI, of a two year doubling in the rate of acceleration. Based on that we can predict that around 2022 the doubling rate of AI power will be about one month.
Still very few data points, and vague, rather than precise predictions. However what matters is the principle of jolting: an increasing rate of acceleration characterizing artificial intelligence applications.
What will be the consequences of this?