AlphaCode, code generator from Deepmind, reviewed on Codeforces

DeepMind, the subsidiary of Alphabet (formerly Google), specializing in AI, presented AlphaCode, a tool capable of writing computer code completely autonomously to solve problems requiring logical thinking and understanding of language. She confronted this new code generation system with human developers during programming competitions on the Codeforces platform where it obtained a score of 54.3%.

DeepMind Technologies, a British company specializing in artificial intelligence, created in 2010 by Demis Hassabis, Mustafa Suleyman and Shane Legg was bought in 2014 by Google for more than 628 million dollars. it aims to findthe best techniques from machine learning and systems neuroscience to build powerful general purpose learning algorithms”. Renowned for its Alphago program which beat several go champions before winning a definitive victory over the world No. 1 in 2017 or AlphaFold, capable of predicting the structure of proteins from their amino acid sequences, DeepMind works on the reasoning rational and predictive, as well as deep learning and deep reinforcement learning type learning methods.

Competitive programming with AlphaCode

The development of machine learning models similar to AlphaCode makes it possible to generate code from algorithms. Several companies have created tools for self-completion of code: Microsoft and Open AI have teamed up to produce Github Copilot, based on Codex, TabNine gave its name to its own. The objective is to enable developers to increase productivity in an increasingly complex activity. It is precisely to bringsolutions competitive programming problems that require an understanding of algorithms and complex natural language» that Deepmind created AlphaCode.

AlphaCode uses large-scale, efficient-to-sample transformer-based language models to generate large-scale code and then filter down to a small set of programs. Deepmind evaluated and validated the performance of its model during competitions on Codeforces, a popular platform where tens of thousands of developers from around the world come to test their coding skills. AlphaCode faced 5,000 competitors in 10 contests, it scored 54.3% and placed around the level of the median competitor, marking the first time that an AI code generation system has achieved a competitive level of performance in programming competitions. Petr Mitrichchev, software engineer, google and world-class competitive programmer said:

“Solving competitive programming problems is a very difficult thing to do, requiring both good coding skills and creative problem-solving in humans. I was very impressed that AlphaCode could make progress in this area and delighted to see how the model uses its understanding of utterances to produce code and guide its random exploration to create solutions.»

Mike Mirzayanov, founder of Codeforces, added:

“I can safely say that the results from AlphaCode exceeded my expectations. I was skeptical because even in simple competitive problems you often have to not only implement the algorithm, but also (and this is the hardest part) invent it. AlphaCode managed to perform at the level of a promising new competitor. I can’t wait to see what awaits me!»

An AI in its infancy

According to DeepMind, this AI is still in its infancy but “Our exploration of code generation leaves plenty of room for improvement and hints at even more exciting ideas that could help programmers improve their productivity and open the door to people who don’t currently write code. We will continue this exploration…”.

For now, AlphaCode has been evaluated in competitive programming but does not yet have the skills of a human developer. On the other hand, like other code generators, it is based on existing codes, so could be accused of plagiarism.

We would love to say thanks to the writer of this post for this outstanding web content

AlphaCode, code generator from Deepmind, reviewed on Codeforces


Visit our social media profiles and also other related pageshttps://www.ai-magazine.com/related-pages/