Some believe the proliferation of artificial intelligence and robotics is putting our privacy, our jobs, and even our safety at risk. More and more tasks go to silicon-based brains. But even the most vocal critics cannot fail to recognize the obvious benefits that AI and automated systems are preparing for humanity. As part of the Grand Challenges project, the BBC brought together experts who set out their vision for the future in the presence of machines and artificial intelligence.
“We should not view AI as something that competes with us, but as something that can enhance our own abilities,” says Takeo Kanade, professor of robotics at Carnegie Mellon University. Because AI has a tolerance for boredom, and is also able to identify patterns much better and faster than humans. Automation has already begun unraveling the world's most complex knots, from disease to cruelty.
And it can make our lives safer in the 21st century.
Fighting infectious diseases
For billions of people around the world, the buzzing of mosquitoes near their ears can mean much more than an annoying bite - it can be a harbinger of illness and even death. One species, Aedes aegypti, has especially spread from Africa to nearly all tropical and subtropical regions, carrying Dengue fever, yellow fever, Zika and chikungunya (a virus that causes crippling joint pain). Dengue alone infects 390 million people in 128 countries every year.
"This mosquito is a tiny demon," says Rainier Mallol, a computer engineer in the Dominican Republic, a hot spot for Zika. Together with Desi Raja, a medic from Malaysia (another country at risk of contracting the virus), the couple have developed AI algorithms that predict where outbreaks are most likely to occur.
Promotional video:
Microsoft's Project Premonition Uses Drones to Find Pathogens in Zika Hot Spots
Their Artificial Intelligence in Medical Epidemiology (Aime) is a system that combines the time and location of each new Dengue case reported by local hospitals with 274 other variables such as wind direction, humidity, temperature, population density, housing type. “These are all factors that determine the spread of mosquitoes,” explains Mallall.
Tests in Malaysia and Brazil have shown they can predict outbreaks with an accuracy of about 88% in three months. The system also helps locate the epicenter of an outbreak to within 400 meters, allowing local medics to intervene in time with insecticides and bite protection for local residents.
Aime is also evolving to predict Zika and Chikungunya outbreaks. Huge tech companies are taking this idea in their own way: Microsoft's Project Premonition, for example, uses autonomous drones to detect mosquito pockets, and uses carbon dioxide and light traps to catch mosquitoes. The DNA of the mosquitoes and the animals they bite is then analyzed by machine algorithms that uncover patterns in gigantic amounts of data better and better each time - and find pathogens.
Weapon Fight
Over the past year, 15,000 people have died in the United States due to gunfire. This country has the highest rate of gun-related violence in the entire developed world. To tackle the problems of indiscriminate shooting and gun-related crime, some cities across the country are turning to technology for help.
An automated system that hears gunfire sounds through a series of sensors can be used to locate where shots were fired and alert security forces within 45 seconds after the trigger was pulled. ShotSpotter uses 15-20 acoustic sensors per square kilometer to detect the characteristic “pop” of a shot, locating its birthplace with an accuracy of 25 meters.
Machine learning technologies are used to confirm that the sound was a gunshot and count the number of shots fired to indicate whether the police will deal with a lone gunman or multiple criminals, and whether they are using machine guns or not.
Already 90 cities - mostly in the US, but also in South Africa and South America - use ShotSpotter. Small systems have also been deployed across nine U. S. campuses in response to the recent campus gunfire.
Ralph Clarke, CEO of ShotSpotter, believes that in the future, this system can be used for more than simple incident response.
“We are looking to understand how our data can be used for predictive capabilities of police officers,” he says. "Machine learning can be combined with weather, traffic and more to inform police patrols more accurately."
Fighting hunger
About 800 million people worldwide rely on cassava (cassava) roots as their main source of carbohydrates. This starchy yam-like vegetable is eaten like a potato; it can also be ground into flour for making breads and baked goods. It can grow where other crops cannot, making cassava the sixth largest food plant in the world. However, this woody shrub is also vulnerable to disease and pests, which can devastate entire vegetable fields.
Researchers at Makerere University in Kampala, Uganda have teamed up with plant disease experts to develop an automated system aimed at fighting cassava disease. The Mcrops project allows local farmers to photograph their plants with cheap smartphones and use computer vision to detect signs of four major diseases that are devastating cassava crops.
“Some of these diseases are extremely difficult to recognize and require different actions,” explains Ernest Mwebase, a computer scientist who leads the project. "We give farmers a pocket expert so they know whether to pollinate their crops or destroy and plant something else."
This system diagnoses cassava diseases with 88 percent accuracy. Typically, farmers need to call government experts to visit farms to identify diseases, which takes days and weeks for the disease to spread.
Mcrops also allows you to upload snapshots to a database, which is then used to diagnose outbreaks. Mwebaze hopes the technology will also automatically detect problems with other plant species, such as bananas.
Fighting cancer and vision loss
Cancer causes more than 8.8 million deaths worldwide, and 14 million people are diagnosed with some form of cancer each year. Early detection of cancer can significantly increase a person's chances of survival and reduce the risk of recurrence. Screening is one of the key ways to detect cancer early, but it is very, very difficult and time consuming to understand scans and other test results.
Google's DeepMind Can Help Physicians With Cancer Treatment With Machine Learning To Help It Identify Healthy Areas Of A Patient's Tissue
DeepMind and IBM are applying their AI technologies to this problem. DeepMind has teamed up with UK NHS doctors at University Colleges in London to train its AI-based program to treat cancer by separating areas of healthy tissue from tumors in head and neck scans. She also works with Moorfields Eye Hospital in London, detecting early signs of vision loss on eye scans.
“Our algorithms are capable of interpreting visual information from scans,” says Dominic King, Clinical Chief at DeepMind Health. “The system learns to identify potential problems and recommend the correct course of action to the doctor. It is too early to comment on the results, but they are already very encouraging."
King says AI techniques can help doctors make diagnoses faster by sifting through scans and prioritizing those that are recommended for immediate consideration.
IBM also recently announced that Watson's AI can analyze images and evaluate patient records, pinpointing a tumor 96% of the time. The system is now undergoing medical trials in 55 hospitals around the world, helping diagnose breast, lung, colorectal, cervical, ovarian, stomach and prostate cancers.
Without turning off the light
Amid a heated debate over whether climate change could have triggered two catastrophic hurricanes on a historic scale in the United States, how could artificial intelligence be maximized to research the use of clean, renewable energy to prevent further damage that leads to climate problems?
People around the world are increasingly relying on renewable energy sources to combat climate change and pollution caused by fossil fuels, and the task of balancing energy grids with such intermittent sources is becoming increasingly difficult. The proliferation of smart meters - digital energy monitors that automatically record consumption - will also provide a lot of data on how and when consumers use energy. The European Union alone plans to install 500 million smart meters in homes by 2020.
“Managing all of these assets is impossible for humans because response times are often on the order of a few seconds,” says Valentin Robu, assistant professor of intelligent systems at Heriot Watt University in Edinburgh. He is working with the UK company Upside Energy to develop new ways to manage power grids.
They create machine learning algorithms to monitor production and energy demand in real time. What does it mean? That energy will be stored during quiet hours and then released during peak hours, for example in the morning, when everyone wants to make their own coffee. As electric vehicles and home batteries become more prevalent, technology can be used to store energy and evenly distribute renewable flows.
Robu also says AI can be used at an even more basic level, helping to reduce our demand for connected devices. For example, refrigerators can be controlled directly by AI so that they only turn on when the demand for electricity is at its lowest on the grid.
Ilya Khel
