How is AI applied in nature?
The UK will use AI to predict the movement of clouds
In the UK, they want to increase the efficiency of solar panels to generate environmentally friendly energy. To do this, we created a machine learning model that will predict the movement of clouds with an accuracy of up to a minute. The worker of the national grid of the UK National Grid ESO has signed an agreement with the non-profit organization Open Climate Fix to form an AI-based tracking system that compares the actions of clouds with the precise location of solar panels. The goal is to predict the impact of this movement on the production of solar energy.
The software will predict the movement of clouds in hours and minutes, not days. The hi-tech can increase the accuracy of solar activity forecasts by up to 50%.National Grid ESO is responsible for handling the balance of supply and demand in the UK electricity grid. As the company explains, the unpredictable nature of the sun and wind seriously complicates the task. To solve the problem, Open Climate Fix has taught a machine learning prototype to analyze satellite pictures.
The project, which began in August, is designed for 18 months. The amount of funding is $683,100.
Pizza Hut is preparing an AI that will recommend pizza “according to the weather”
Fast food chain Pizza Hut is developing a proprietary AI system that will tell customers which pizza to choose based on the local weather. The machine learning algorithm will be ” fed ” data about customer orders along with some information about them, their location and the weather forecast at this point. Based on this, the optimal pizza for a particular client will be selected here and now. This way, the company expects to get more satisfied visitors and increase sales.
So far, this system is at an early stage of development. It is not specified exactly how the weather will affect the decision of the neural network.
The project is being handled by the IT department of Digital Ventures, which appeared in Pizza Hut in 2016. The specialists decided not to use ready-made AI programs, but to create their own technologies tailored to the specifics of the business.
Pizza Hut is not the only fast food restaurant chain implementing AI. In 2019, McDonald’s bought the startup Apprente, which specializes in voice recognition. Its developments were planned to be used to speed up the customer service of MakAvto.
The world’s first AI ship ended its oceanic voyage with a breakdown
The autonomous AI vessel Mayflower (MAS), which is supposed to recreate the historical journey of British immigrants across the Atlantic, is returning to the UK after a breakdown. The ship set off on a voyage of 5,630 km from the British Plymouth to the American Massachusetts last Tuesday. The trip was expected to take about three weeks.
It was planned that the ship would not only recreate the legendary route of the British pilgrims, but also help researchers conduct experiments with collecting data on marine flora and fauna, as well as take samples of plastic waste.
The 15-meter-long solar-powered trimaran is capable of reaching speeds of up to 20 km/h and is controlled by an onboard AI that receives information from six cameras and 50 sensors.
Mayflower’s problems began three days after going to sea.
Users who followed the movement of the ship from its online panel on the network became worried when the streaming video was turned off. The Mayflower Twitter account confirmed that “secondary systems” are temporarily disabled to save energy.
According to the developers, there was a “small mechanical problem that can be found on any boat”. However, the experiment was difficult to continue due to the sharply dropped speed of the ship-the AI ship was given the command to return to Britain.
Estonian researchers have taught AI to generate human DNA
Artificial intelligence can almost perfectly draw non-existent faces, legs, cats, resumes, anime characters and much more. Specialists of the University of Tartu in Estonia have trained an algorithm to generate deepfakes of people at the molecular level: it creates unique sequences of the human genome, writes The Next Web.”Generative-adversarial neural networks have been effectively used in many areas for the last 10 years, including for creating photorealistic images. We are using a similar approach to genetic information to automatically study its structure and, for the first time, to generate high-quality, realistic genomes, ” the scientists say.
“Artificial genomes” are created on the basis of real human ones and are indistinguishable from them-with the exception that they are completely synthesized by the model. They preserve many complex characteristics of the original genomes and can be used in further research.
As TNW notes, this solves the problematic issue of data privacy for geneticists and protecting the privacy of the people to whom they belong. Public access to DNA datasets is limited, and the procedure for obtaining it is complex and long. The Estonian ML system can become an alternative to such databases: this will eliminate the lack of necessary data and advance genetic research.
Harvard has created an AI to predict how drugs affect life expectancy in mice
Scientists from Harvard University have presented an AI system that predicts how various ways of prolonging life in mice work. According to the researchers, in the future it can be used to create appropriate procedures and medicines for humans, writes The Next Web. The researchers conducted observations of 60 mice until the moment of natural death. Throughout their lives, scientists measured their ability to walk, spinal curvature and hearing condition. Based on these data, two AI models were built: the first determines the biological age of the mouse, and the second-how much longer it will live. The forecasts turned out to be correct with an accuracy of up to two months.
The calculations of the system are planned to be used for rapid testing of various measures that increase the life span of mice, and then-to start research on humans.
The mice were divided into two groups: they were either given special drugs or prescribed a special diet. Artificial intelligence was able to accurately predict how much these measures will slow down the aging of experimental subjects, the researchers say.
It takes up to three years to study the effectiveness of a drug or diet. Predictive biometrics can speed up such studies.
In addition, AI has found that some processes in the body are more accurate indicators of future health: for example, hearing loss and tremor are more strongly associated with biological age than vision or moustache loss.
The new predictive system is not yet used in humans, because in the case of a person, much more parameters that affect the life span need to be taken into account. Scientists also want to collect a suitable dataset with information on people who were observed at the age of 60 to 90 years, and mortality among them.
The researchers plan to create a tool for predicting the effectiveness of various ways to increase life expectancy in humans.