Google and the AI that creates and selects algorithms that evolve without human intervention.

Hello dear readers!

Artificial intelligence is evolving, it is advancing by leaps and bounds, so much so that when I read about the advances in AI I do not know whether to rejoice or scare me, and that is that man continues his career trying to replicate and overcome human behaviors and actions, seeking to overcome the capabilities of the human being.

Source

Engineers from the Google Brain division, claims to have developed an artificial intelligence based on Darwin's theory, which uses evolution and survival of the fittest.

"While most people were taking baby steps, they took a giant leap into the unknown," says Risto Miikkulainen, a computer scientist at the University of Texas, Austin.

AutoML-Zero (name given to this new version) follows a selection process similar to that proposed by Charles Darwin in his Theory of Evolution.

The engineers explain in the published document everything you need to know about it:

"Human-designed components skew search results in favor of human-designed algorithms, possibly reducing AutoML's innovation potential"

AutoML-Zero randomly picks 100 candidate algorithms and tests them with relatively simple tasks. The best performing algorithms go to a next phase where their code is modified slightly and randomly and then tested again.

Source

At the same time, variants that have been found to be effective are added to the rest of the algorithms. And thus phase after phase until we have a more varied "population" of algorithms and with better capacities to solve the proposed tasks.

This performance follows a pattern similar to that of the survival of the fittest or fittest.

“Our ultimate goal is to actually develop novel machine learning concepts that even researchers could not find,” he says So Quoc Le, a computer scientist at Google.

Now, we must ask ourselves where this technology that practically takes the human being out of the game will take us, logically according to the engineers, there is still a lot of work to be done but they have made a very big leap, a significant advance in the development of this technology.

Let me know your opinion ...

The fact that artificial intelligence can evolve to "improve" without human intervention is something safe for humanity?

Are we advancing to the time of the true autonomy of the machines?

Source consulted: ScienceMag.org