Seen on Apple News This amazing article About the weaknesses of artificial intelligence in mathematics. Artificial Intelligence (AI) systems like ChatGPT still perform poorly in mathematics.
In short: AI systems like ChatGPT are developed based on large language models (GML) and not on mathematical models. In other words, we must clearly distinguish that mathematics and arithmetic or computer processing are far from identical. AI systems use their computing power to determine probabilities and determine which words to choose from their huge database to build a meaningful sentence. While mathematics is based on inference, pure logic.
The “probability” is that 2 + 2 = 4
Based on probability, AI is a model that does not take into account the concepts of “right” or “wrong,” as in mathematics.
So, to determine that 2 + 2 = 4, AI models rely on probability. So, to arrive at the number 4, the AI model will not think like us humans, but will search in its huge data bank that the string of numbers and signs in “2 + 2” is the number 4.
It doesn’t happen all the time to use the mathematics reasoning logic, as the reasoning techniques introduce the concepts, the numbers, the ensembles and the figures in the mathematics that remain in the older bastions that are resistant to your problems. ‘Artificial intelligence.
For us, it is easy to design a triangle, but for an AI machine, it is more difficult.
In December 2022, on the occasion of its launch, many quickly discovered the limitations of ChatGPT which had to answer a simple mathematical problem like this: “If five machines produce five things in five minutes, how long will it take? Does it take 100 minutes?” Machines to produce 100 pieces? ChatGPT responded 100 minutes ago. forged! The answer to this question, which is fundamental to a mathematician, is five.
Note: To students who have done their math homework using online AI tools, you may be disappointed!
