This article looks at how artificial intelligence can help expect the hourly consolidation of air toxin Sulphur ozone, element matter (PM2.5), and Sulphur dioxide. As one of the most excellently procedures, AI can efficiently prepare a model on a large amount of data by using large-scale streamlining computations. Even though several works use AI to predict air quality, most of the earlier studies are limited to long-term data and easily instruct regular relapse designs (direct or nonlinear) to expect the hourly air pollution focus. This paper suggests advanced analysis to simulate the hourly environmental change focus based on previous days' weather-related data by calculating the expectation for more than 24 hours as an execute multiple tasks learning (MTL) issue. This allows us to choose a suitable model with a variety of regularization strategies. We suggest a useful regularization that maintains the assumption patterns of concurrent hours to be nearby to each other, and we evaluate it to a few common MTL expect completion such as normal Frobenius standard regularization, normal atomic regularization, and '2,1-standard regularization. Our tests revealed that the suggested boundary declining concepts and constant hour-related regularizations outperform open product relapse models and regularizations in terms of execution.