Facial Recognition Technology: How intelligent is artificial intelligence in identifying black and Asian faces?
According to a US government survey, algorithms that identify human faces are far poorer in identifying the faces of blacks and Asians than of white-race faces.
Government research also shows that women are more likely to be mistaken in identifying women than men of the black race.
Further research raises further doubts as to whether law enforcement agencies should use this type of technology.
Demands ban on emotion testing technology
China: Identification of accused in crowd of 60 thousand
'The sheep can recognize human faces'
The National Institute of Standards and Technology (NUST) has tested 189 algorithms of 99 companies, including Intel, Microsoft, Toshiba and Chinese companies Tencent and DDChousing.
Amazon sells its face recognition technology 'Recognition' to US police forces, but the company did not submit its own sample for the study.
'One to one matching'
The company previously described a Massachusetts Institute of Technology study as 'misleading'. The report said that the recognition performance was particularly poor in identifying the faces of black women.
According to the report, when a particular image was matched to another image of the same person (one-to-one matching or matched one-to-one), many artificial intelligence algorithms saw black and Asian faces ranging from 10 to 100. More than once Caucasian faces misidentified.
And in one-to-one matching, there is a greater likelihood of misidentifying black women, where a particular image is likened to the many faces in the database.
Beni Thomson, a member of Congress and chairman of the US House Committee on Homeland Security, told Reuters: "The US administration must re-evaluate in the light of the new terrorist consequences of facial recognition technology."
Computer scientist and founder of the Algorithmic Justice League, Jawa Bolamoni, has "vehemently denied" the report, claiming that bias is not a problem in artificial intelligence software.
The NIST study examined two types of errors in algorithms:
- False positives, in which the software incorrectly calls two different people's pictures as one person's image
- False negatives, in which the software makes a mistake in reaching two pictures of the same person.
Database images provided by the Foreign Ministry, Homeland Security Department and the FBI were used to test the software. It did not use any surveillance images via social media or video.
"It would be wrong to say all the algorithms," said Patrick Grother, a computer scientist at Nest and the lead author of the report. But in most of the algorithms we studied, we found empirical evidence about population differences.
SenseTime, a Chinese company whose algorithm was found to be flawed, said it was the result of "bugs" and has now been corrected.
A spokesperson told the BBC: 'These results do not translate our products because they are in the testing stage before being released to the market. This is why all our commercial solutions have the highest level of accuracy. '
Various US cities, including San Francisco, Oakland, California, Summerville, and Massachusetts, have banned face recognition technology.
Indian Space Agencies ISRO and DRDO have released the mission Shakti Shakti, in which a satellite was struck 300 km from a missile.India successfully tested a satellite missile (AAT) on March 27, and Prime Minister Narendra Modi announced that India had become the fourth country to have this capability.
Is India's 'Mission Shakti' a danger to itself?
China and the United States have expressed concern over the debris originating from India's experience where the ruling government is proud.
During this experiment, Indian scientists targeted their own satellite located in orbit near the Earth, disintegrating 700 kilograms of microstar satellites 300 km away, resulting in debris in space. Gaia.
US Secretary of Defense Patrick Shanahan called on countries experiencing satellite weaponry, saying they should not risk collecting debris in space.
According to the news agency Reuters, he addressed the US Army Southern Command in Florida and said, "My message would be that we all live in space and we should not spread dirt there." Space should be the place where we do our business. Be a place where people have the freedom to do their work. '
"Eye in Space", India launches a spy satellite
New record in India's space mission
When the BBC inquired with former defense affairs expert Komodo Ade Bhaskar about whether the United States was right in its warning, he said that no one should have the right to spread rubbish in space.
Prime Minister Modi expressed India's new power in addressing the nation
Pieces larger than ten centimeters, called debris, are about 34 thousand and smaller than ten centimeters and fragments larger than one centimeter are about one million, while smaller fragments are about 12 crores and are used to test India. Prior to that, there was a spread of responsibility for the countries before India, including the United States, Russia and China. "
We asked them if these debris could even come towards the earth, so they said that there are different levels of orbit of the earth. Sometimes they can even approach the earth if they are closer to the ground and its forces fall in the attraction, but if they are too far away, they usually disperse into space. '
He said there are usually three orbits in the satellite. 'There is a low orbit that is between two thousand kilometers from the Earth's atmosphere. The second is the area of the central orbit, which is between 2,000 and 20,000 km away from the earth, while the uppermost 36,000 kilometers are rotating geosynchronous satellites and if they disappear or break for some reason. So they roam there. '
He said that what India has experienced is 300 km from Lo Orbit and that the debris of this orbit is dispersed further in ten days and gradually comes to land and into the sea or land. Falls off
Former Commodore Ade Bhaskar can be seen meeting the former president and scientist APJ Abdul Kalam
We asked the same question to renowned poet and poet Gohar Raza. He said, 'Here it is important to understand that this experience is a deliberately smashing experience as India did. It hits the target and is broken into pieces and turned into debris. Now after that no computer in the world can tell where the pieces will go. The speed of these pieces will vary and no one can predict in which direction. So, scientifically, I find the claim wrong that the debris will be gone. Mobs are stopped because of the attraction on earth, but in space they have always been moving. '
In one of his articles, Pablo Bagla, a journalist writing on scientific issues, expressed the fear that when India launches a PSLV satellite off its east coast on April 1, these debris may collide. Or is it that India's 'mission shakti' has not become a threat to its own satellite?
In response, scientists said that this could happen in principle, but there are already a lot of debris floating in space, and with each satellite launch it is increasing.
Gohar Raza said, 'The debris that is being created in space is moving very fast and it can collide with anything. They can also collide with their satellite or from other countries 'satellites, or they can also hit missiles that we are sending.'
The Prime Minister of India had claimed with the announcement of 'Mission Shakti' that India had become more secure with this experience.
Gohar Raza, a scientist, said, "I don't think such experiments should be done." This does not save a country. Frequently the word is used that we are saved, but I think when we develop such a technology, we endanger the whole humanity instead of being safe. '
If such experiences are dangerous, then is there an international convention or agreement?