Column|Artificial intelligence was to win the gold medal at the Junior Mathematical Olympiad.
Olympics didn’t go wrong. Not from Finns, and especially not from artificial intelligence.
In July, two young Finns received medals at the Mathematics Olympiad in Bath, England. Aino Aulanko brought silver and Zhongyi Li bronze.
The name of the competition is a bit lofty. However, the Mathematics Olympiad is the world’s most prestigious event for talented mathematicians under the age of 20.
Medals were also distributed generously: 58 gold, 123 silver and 145 bronze. There were 609 high school students from 108 countries.
Artificial intelligence programs took the spotlight. Search giant Google’s Deepmind programs Alphaproof and Alphageometry2 won silver. Geometry missed the gold by one point.
Mathematics is the kingdom of reasoning and logic. Few believed in these first timers.
Artificial intelligence trained ferociously for the Olympics. They solved and proved millions of mathematical problems.
It seems strongly that artificial intelligence can solve demanding mathematics.
However, Alphageometry2 was allowed to use days to solve the tasks. The time of young talent was limited to hours.
One of the developers of Deepmind artificial intelligence Alex Davies described the achievements as a breakthrough.
The artificial intelligences seemed to reason sensibly and with high quality – and all by themselves. No confusion, no hallucination.
Although Davies represented the voice of the owner of artificial intelligence, the search giant Google. That’s why the newspaper The New York Times asked other mathematicians about the value of the medals.
“
Mathematics is the kingdom of reasoning and logic.
One of them was a programmer and mathematician Joseph Myers. He was impressed. He knows what he’s talking about. Myers won Olympic medals at a young age.
In the hunt for medals, Alphageometry2, as its name suggests, succeeded precisely in geometry. Another AI, Alphaproof, focused on proof.
Artificial intelligence researchers say artificial intelligence still has a long way to go to research-level mathematics.
In the Olympics, you have to know “only” solution techniques. It is not necessarily related to the study of mathematics.
However, artificial intelligence is slowly coming to the researcher’s aid. They can speed up the proof.
“I have a colleague now. It costs a few dollars a month. I can discuss ideas with it,” wrote one mathematician in the NY Times.
Another reader ponders that maybe we don’t need an AI that solves math. We need artificial intelligence that can teach people how to think.
The author is the editor of HS science editorial.
#Column #Artificial #intelligence #Olympics