In this final video, I was tempted to 
make some predictions about the future of 
research on neural networks. 
Instead, I'm going to explain to you why 
it would be extremely foolish to try to 
make any long term predictions. 
I'm going to try and explain why we can't 
predict the long term future by using an 
analogy. 
Imagine you're driving a car at night, 
and you're looking at the taillights of 
the car in front. 
The number of photons that you receive 
from the taillights of the car in front 
falls off at one over d squared, where d 
is the distance to the car in front. 
That's assuming that the air is clear but 
now suppose there's fog. 
Over short ranges, the number of photons 
you get from the tail lights in front of 
you, still falls off as one over d^2. 
Because over a short range, the fog 
hardly absorbs any light. 
But for large distances, it falls off as 
E to the -D. 
And that's because fog has an exponential 
effect. 
Fog absorbs a certain fraction of the 
photons per unit distance. 
So for small distances, fog looks very 
transparent, but for large distances, it 
looks very opaque. 
So, the car in front of us becomes 
completely invisible at a distance at 
which our short range model, the one of a 
discreet model, predicts it will be very 
visible. 
That causes people to drive into the back 
of cars in fog. 
It kills people. 
The development of technology is also 
typically exponential. 
So over the short term, things appear to 
change fairly slowly. 
And it's easy to predict progress. 
All of us, for example, can probably make 
quite good guesses about what will be in 
the iPhone six. 
But in the longer run, our perception of 
the future hits a wall, just like with 
fog. 
So the long term future of machine 
learning in neural nets is really a total 
mystery. 
We have no idea what's going to happen in 
30 years' time. 
There's just no way to predict it from 
what we know now. 
Because we're going to get exponential 
progress. 
In the short run however, in a period of 
say three to ten years, we can predict it 
fairly well. 
And it seems obvious to me that over the 
next five years or so. 
Big deep neural networks, you are going 
to do amazing things. 
I'd like to congratulate all of you who 
stuck it out long enough to get this far. 
I hope you've enjoyed the course and good 
luck with the final test.