Chatgpt threatens to end millions of jobs in the coming years if the situation does not change. The programmers are in the spotlight, but Not even Openai’s Chatbot himself is safe in the race to dominate in AI.
Sam Altman’s company has a new competitor: a programming agent that not only follows patterns, It can generate sequences from scratch, even literally rewrite your own code. Darwin-Gödel (DGM) has led the learning of AI to the next level.
An AI that learns itself code
The first Darwin-Gödel prototype was tested in May of this year. Researchers behind this AI wanted to create a self -manageable programming agent, but in a somewhat different way from Chatgpt and other chatbots.
Darwin-Gödel not only becomes smarter with use, he is also able to modify his own structure to offer better answers. The AI tool Save codes that work to use them in the future, add variations and correct failures.
This programming agent does not require an expensive resentment process in which to add more information and examples, does not even need a human professional to make adjustments in the code.
Darwin-Gödel has ended the limits of the AI so far. Artificial intelligence could only execute the actions requested by users, now it can radically change its operation without any programmer modifying the code.
This model performs automatic updates, you can read your own files in Python, try processes and put them into practice if you have verified that they work. Darwin-Gödel Use the formal test method with measurable results instead of the classic test and error much slower and expensive.
Final goodbye to programmers
The programmers had a complicated situation since the arrival of Codex for Chatgpt and other coding tools, now it has only worsened. Darwin-Gödel is totally autonomous.
The work of programmers would change forever with the arrival of AI. The programming agents would be in charge of making the code, and a professional reviews it and adds the necessary corrections.
Darwin-Gödel totally dispenses with humans. The machine can add improvements to its code and modify its own structure if it fails to find an answer, so the contribution of programmers is no longer necessary.
Darwin-Gödel’s results speak for themselves
Researchers have tested DGM capabilities in the SWE test bank, a parameter to measure automated code repair. The first results have exceeded all expectations.
Darwin-Gödel began to generate code with a 20%success rate. After a continuous use of the programming agent, the success number has been fired at 50%more than double the initial value.
The AI agent has not needed additional training nor has he had external help, he has simply improved himself. Each version of Darwin-Gödel is smarter and more efficient than the previous one.
Know How we work in NoticiasVE.
Tags: Artificial intelligence, programming