In other words, the scale of AI pornography detection previously used by major internet companies, which involved hundreds of people, has now been reduced by another 90%.

AI-based content moderation was just the first practical application of the M-language.

The AI ​​technology it represents has seen a significant improvement in both computing power and the difficulty of implementing the algorithms themselves.

It allows for the easy construction of most originally complex architectures and can be widely used in the industrial field.

A large number of easily combinable modular components, perfect fit for writing your own layer types, computation graph abstraction, data and model parallelism, dynamic type stability, and more.

After algorithm engineers used it, projects related to the M language on GitHub became popular in the community.

In overseas programmer communities, active experts are complaining, and even some rarely speaking experts have started to chime in:

"Merlin's technology far exceeds our imagination. If he developed the M language himself, then his expertise in the field of AI surpasses all of ours."

It's really hard to find an AI programming language that's completely flawless.

Caffe can port the fast convolutional network implementation of MAB to C and C++, making it suitable for feedforward networks and image processing, and models can be trained without any code.

However, Caffe cannot be used with recurrent networks and has extremely poor scalability.

Google's TensorFlow framework can generate a computation graph and then perform automatic differentiation, eliminating the need for manual coding when trying new neural network arrangements.

However, it runs very slowly and is very prone to errors in large software projects.

Basically, all the AI ​​frameworks launched by major internet giants in recent years have had some problems.

Of course, this is unavoidable; nothing is perfect.

Just like Java, no matter how many people use it or how long it has been active, its shortcomings can still be found.

However, in the field of artificial intelligence, the M-language seems to have no flaws.

It seems to have undergone countless tests and optimizations, reaching a state of perfection.

This makes me seriously doubt that the artificial intelligence Merlin possesses is strong artificial intelligence.

He can realize almost all of the Earth's fantasies about artificial intelligence.

Including self-programming.

It's not so surprising that one piece of code can compile another.

After all, current AI models can already perform simple programming on their own.