The Transformer Neural Networks Saga is Just Getting Started
There are applications that we cannot even dream about!
I’ve been reading many analyses of Transformer Neural Networks. Most of them are missing the fundamental point.
We understand natural language. The computer does not have innate consciousness to do so.
A GPT-3 transformer is an advanced mathematical model that are just beginning to change the way everything operates.
This is the architecture of an GPT-3 TNN(Transformer Neural Network — my personal nomenclature) given below:
The Take Away
We see the input and the output. We see that it processs languages. So we say, it understands languages and listens to them accurately and produces coherent output.
All that is fine.
But the computer is unaware that this is a language.
The AI would have to check a dictionary to grasp the meaning of the word.
To the GPT-3 system, everything is mathematical calculation.
And that is what very few people understand.
If you want to see how GPT-3 works practically, please read this excellent article given below:
The GPT-3 Architecture, on a Napkin
There are so many brilliant posts on GPT-3, demonstrating what it can do, pondering its consequences, vizualizing how…
Many speakers and writers talk about false news, disinformation, photo-shopping, and other such common terms. But GPT-3 goes far beyond all that.
All the big FAMGA companies are into TNNs in a very big way. AlphaCode, AlphaCode2, and many other examples of enormous potential exist in TNN domain space already. But this is just scratching the surface.
The computer does not know what the data is.
Yet it produces the correct output.
After all, AI is a hybrid of optimization and statistics for interpretation.
That means it can be applied to any data anywhere which requires prediction, modeling, and forecasting.
- Stock Market Forecasting and Predictions
- Crypto Market Forecasting and Predictions
- Perfect Weather Predictions (at a fraction of the cost)
- Programming language to programming language Automatic Porting to different platforms
Let’s go even further.
- Formation of the Universe
- Proving or Disproving the Existence of the Multiverse
- How to build truly intelligent robots (I am, because I think)
- Accuracy of Paleontology
- More Information on the Higgs boson.
Why stop there?
- Cure for every single type of cancer
- Zero-shot learning to reduce costs.
- A Vaccine for Malaria and one for AIDS
- Extremely deep layers of learning in TNNs (as GPU costs decrease)
- The Original Simulation of how life on earth was formed
- Are there aliens in the Universe?
We have limited data. The TNNs convert from small data inputs to huge, massively generated possible solutions. Scientific progress on Earth could easily increase by a factor of 10x!
By creating the right transformers with the right data, we can solve any problem we want.
By using a possible digital encoding, every human issue could be solved.
The computer is not intelligent (yet). It’s a machine. It handles binary data.
The encoding and the choice of problem formulation could result in huge progress in human understanding,
We have so much to extrapolate from the data we have already that we earlier had to manually label and describe but now even that can be done automatically.
And the work will lead to new discoveries.
Stunning new discoveries.
I hope you’re on the front line when it happens.
And I guess nobody needs to be told to venture into this particular field.
The potential rewards are too high to resist.
Wrapping Things Up
A human brain is a wondrous tool.
Now, we are more powerful than at any time in the past.
What will happen in the next few years?
I can’t wait to find out!
All the best.
You are on Earth for a purpose.
And I pray that this article gives you exactly that.
Peace of Christ!
Welcome to the new world! And it is just awesome and magical.