Stanford University's 2019 AI Index report indicates that the speed in the evolution of AI (Artificial Intelligence) by far outpaces Moore's Law.
We're pretty familiar that processor capacity doubles every 18 to 24 months - meaning application performance progressively improves with every cycle. While this has shown exponential results in the last few decades, this is peanuts compared to what will be happing the next few...well...months.
A report from Stanford University, based on research conducted together with McKinsey & Company, and Google found that Artificial Intelligence computational capacity is accelerating faster than conventional processing capacity evolutions.
Before 2012 the evolution of AI kept pace with Moore's Law - meaning computing capacity doubling roughly every two years. However, post 2012 - this has been happening every 3.4 months.
OK. Let that sink in for a second.
If you thought we have had seen exponential evolution - you ain't seen nothing yet. You better buckle up for the next coming years - because there's a rough ride coming at us with a technology acceleration that our human brain cannot comprehend.
The Stanford research shows, for example, that over 18 months - the time needed to train a neural network of supervised image recognition dropped from three hours in October 2017 to 88 seconds in July 2019.
The reports also looked at something that's called the Resnet image classification model - in plain English - how computers teach themselves to recognize objects. It looked at how long it took AI to achieve a high-level precision. In October 2017, it took 13 days of training to achieve 93% accuracy. The associated cost for processing power would have been 2.323€. The latest benchmark (see Standfords' Dawnbench report) in September 2018 showed that cost dropped to just over 12€.
Major players like Google are even expanding their commitments towards AI development and are aggressively discounting it's available resources for GPU, TPU and local SSD to make AI development accessible to all.
And they're also making it playfull. Even if you're just a rookie - you can start playing around and experimenting with AI using Google's experiment platform (https://experiments.withgoogle.com/collection/ai)
What happened to the last decade(s)?
The level and speed at which AI is democratizing technically, financially and in any other aspects is mind-boggling - and what we can expect is that for the first time since the last industrial revolution will mean an increase in productivity per employee.
After all - despite the digital revolution we've witnessed the last few decades - productivity and margin per employee have barely improved, and definitely not equivalently as technological evolution has evolved.
The next few months, we will see accelerated adoption of narrow AI applications such as creating short term impact on job descriptions and tasks performed by employees and agencies.
Only in the field of Sales & Marketing - I have curated a list of roughly 40 narrow AI tools (ping me if you would want the list). When you map those tools out on four stages of a Marketing process: (I refer to them as 1) Strategize 2) Shape 3) Share and 4)Scale your brand) - the tasks those AI tools will take on look something like this.
All together - what you see above is the job description of your average marketer...that being said - this is spread across roughly 40 tools...
While we're not there yet that AI will replace your job as it is - it will make super humans out of us. I've written a book - Obsessed. Without Grammarly's AI, it would not have been as good as it is today. I DJ at night. Without Algorittms's AI teaching me how to mix - I would have never been able to do so.
I still very often come across many disbelievers of AI. I believe the next decade will embark the rise of a new social classification of individuals and companies. Those that will rise together with the machines. And those that will fight against them.
May the best man or machine win.
Have a growing 2020.
Want to find out more on how to strategize and implement AI? #Getobsessed.