There are hinge points in society which alter things in ways that fundamentally change how society works. The domestication of fire may have been the first. The development of projectile weapons and the efforts of the first farmers are examples. They can be large structural change like the development of the city/state or a knowledge engine like the invention of the printing press. They may be instantaneous like the explosion of the first atomic bomb or aspirational like the Magna Carta, but in all cases society changes as a result.
Recent events such as the invention of the computer or the internet have affected most of you who will read this. Think about the smartphone. That singular device has dramatically changed how we live our lives day to day. Can you remember how it was before everyone had a phone with them at all times? The smartphone was a hinge point that changed society in consequential ways in a single generation.
With AI we are currently living in a hinge point unlike anything society has seen before. It will impact all aspects of society in the next 5 to 10 years in ways large and small that we cannot really predict as we are riders in the rocket and don't know the destination. We know only that the future will be different.
Just a couple of statistics to give you a sense of it. You will hear the politicians and pundits casting clickbait upon the water but you can be confident that they have no idea about what they speak so passionately. Industry will speak the required words of caution as their development teams race forward. To understand the game you need to have a sense of its economics.
Stanford University has studied the cost of AI and machine learning. For a long time it was a toy of research with small wins here and there but a sandbox for the well financed because of cost. But imagine what happens when computer power becomes infinite as a practical matter and virtually free? That is what frees the genie.
The ResNet image classification model is used to assess how long it takes algorithms to achieve a high level of accuracy. In October 2017, 13 days of training time were required to reach just above 93% accuracy. Training an AI-based image classification cost about $2,323 in 2017.
The latest benchmark available on the Stanford DAWNBench running the ResNet model cost just over $12 in September 2018. Imagine that, a nearly 200 times reduction in cost in only one year and that was 5 years ago. It is because of the breakthroughs like that that AI is everywhere today and they only let us see the small pieces.
We have reached a point where a teenager with a slightly indulgent parent and a credit card can explore ideas from his basement that 20 years ago only the government could afford.
In this climate we feel it is important to provide a bit of context to the patient community and we invite you to join us for a webinar about AI and machine learning.
If you’re interested in learning more about AI in a more productive format, join us this Tuesday, July 11 at 11 AM ET in a discussion with some experts out of MIT who have formed a company, “Array Insights”. The webinar is free to attend and will run for about an hour.
Here is the link to register:
Even if you can't attend you should register as we will distribute the recording.
We’re still taking responses for our 2023 patient survey. A large, collective voice from the NAFLD/NASH community is needed to push for meaningful change in our healthcare and patient experience.
Do your part and take 17 minutes to help us along.