Google-owned artificial intelligence company DeepMind has announced a major achievement in competitive computer programming.
After simulating 10 contests, with over 5,000 entrants, the AlphaCode AI system ranked in the top 54% of competitors. There was still work to be done to bring it up to par with high-performance humans, DeepMind said. And his skills might not be immediately applicable to other forms of coding.
Come true The AI doesn’t see the problems in the competitions during training, so solving them requires a combination of critical thinking, logic, algorithms, coding, and natural language understanding.
Typical problems include finding ways to place roads and buildings within certain constraints, or creating strategies to win board games. DeepMind Principal Research Scientist Oriol Vinyals said, “I’ve been fascinated by competition programming all my life.”
Being part of the team that built AlphaCode was “a dream come true”, but he never expected it to reach the “human average among competitors”.
Really difficult Mike Mirzayanov, the founder of the Codeforces platform, which organized the competitions, called AlphaCode a “promising new competitor.”
I can safely say that the results of AlphaCode exceeded my expectations,” he said. I was skeptical because even simple competitive problems often require not only implementing the algorithm but also, and this is the hardest part, inventing it.
Google software engineer and competitive programmer Petr Mitrichev said, “Solving competitive programming problems is a really hard thing to do as it requires good coding skills and creativity to solve problems.”