Thanks to the merger of Google Brain and DeepMind, the new group is developing a new AI model that revolutionizes sorting algorithms. Drawing on its former technologies, the researchers could develop AlphaDev.
How does AlphaDev work?
Inspired by two AI models widely used in video games, AlphaZero and AlphaGo, Google DeepMind researchers could develop AlphaDev.
This new model creates a new generation of sorting algorithms, a necessary technology for a society in perpetual search of computing power and speed. The algorithms created by AlphaDev will be integrated into future Google software.
How are the algorithms generated more powerful?
To create new algorithms, AlphaDev designs them in stages. By comparing them with a pre-established algorithm, the model could identify possible improvements, resulting in a 70% increase in algorithm power.
To design this new generation, AlphaDev uses assembly languages, low-level programming languages. This use has shown the researchers possible avenues of improvement for future AI models.
The new algorithms are already available as open source on the Libc++ library.
This new development will bring major improvements in the scientific and IoT fields. It’s a way of pushing back the limits of our current technologies once again, made possible by Google DeepMind’s data teams. One of the latest technological improvements, also open source, is Falcon, an LLM that surpasses previous models. So, if you’ve enjoyed this article and are considering a career in Data Science, don’t hesitate to check out our articles or training offers on DataScientest.
Source : deepmind.com
