We aim to inform readers not only about developments in the Bitcoin industry, but also about what is happening in related technological fields — cybersecurity, and now the world of artificial intelligence (AI).
ForkLog has gathered the weeks most important AI news.
- In the state of Maine, law enforcement was prohibited from using facial recognition systems.
- In Belgorod Oblast, a pilot project for using AI in courts was launched.
- US automakers must report all crashes involving autonomous systems.
- Independent researchers deemed Tesla’s computer-vision-based Autopilot safe.
- Amazon used algorithms that fired couriers without cause.
- A church-affiliated hospital in Moscow will implement AI technologies for diagnosing diseases.
- Heineken created a robot that follows the user and brings cold beer.
In Maine, authorities ban police from using facial recognition
In the American state of Maine banned facial recognition by law enforcement, with an exception for investigations of serious crimes. In those cases, police would need authorization from the FBI and the states Department of Motor Vehicles.
The law prohibits deploying facial recognition systems in schools and requires law enforcement to maintain a public log of search queries. It also gives citizens the right to sue the state if they believe the government agency used the technology illegally.
The American Civil Liberties Union welcomed the law and called it the strictest in the United States with respect to facial recognition systems.
Artificial intelligence deployed in Russian courts
Russian courts beganuse artificial intelligence to draft documents and verify requisites/details.
A pilot project was launched in Belgorod Oblast, where AI was connected to generate court orders for debt collection related to property, land and transport taxes.
Experts say the technology could ease the workload on government agencies, cut budget costs and minimize errors.
US automakers must report crashes involving autonomous systems
In the United States, carmakers must report crashes involving semi-autonomous and autonomous driving within 24 hours.
Officials from the National Highway Traffic Safety Administration say this will help collect information to improve road safety. Regulators will also require manufacturers to submit monthly reports detailing all incidents with injuries or property damage related to automated driving systems.
Penalties for non-compliance can reach up to $23,000 per day.
The rule covers both Level 2 driver-assistance systems and Level 5 fully autonomous vehicles.
Tesla’s Autopilot based on computer vision deemed safe
Consumer Reportsrestored its ‘Best Buy’ rating for the Tesla Model 3 after IIHS updated its assessment of the camera-based driver-assistance system.
IIHS noted that the Model 3 with Tesla Vision cameras features effective automatic emergency braking and forward-collision warning. These two criteria were central to Consumer Reports’ decision.
In May 2021, the automaker lost recommendations from both organisations after it refused to use radar sensors in its autopilot and focused on developing the computer-vision system. IIHS and Consumer Reports took some time to verify the safety of the updated driver-assistance system before restoring the rating.
Amazon deployed algorithms to fire couriers
Amazon uses algorithms to dismiss drivers in its Flex delivery service.
According to media reports, the system often removes workers without just cause. For example, one driver’s rating fell below the acceptable threshold because the driver punctured a tire and failed to deliver the order on time. Another driver said they also lost a positive rating because they could not access a gated residential complex, and the recipient did not answer calls. In both cases the algorithm deemed it necessary to terminate the drivers.
Participants in the Flex program can appeal the system’s decision, but doing so costs $200. Many drivers say the effort is not worth it.
Nevertheless, within Amazon the program was deemed successful. Since its inception, about 4 million drivers worldwide have downloaded the app.
A church hospital to adopt AI for diagnosing and diagnosing
The Central Clinical Hospital named after Saint Alexius, Metropolitan of Moscow, affiliated with the Russian Orthodox Church, joined the list of pilot institutions for AI deployment.
Under the project, AI is used to improve diagnostic accuracy in radiology and pathology, as well as monitor service quality. The algorithms will expand treatment and diagnostic capabilities for patients with cancer or other diseases, the hospital says.
The hospital also said that they were among the first in Moscow to apply neural networks to speed up COVID-19 diagnosis by analysing chest MRI scans. They plan to expand this direction.
Artificial intelligence trained to predict human actions in video
Researchers at Columbia Universitydeveloped a computer vision technique that predicts a person’s behavior on video several minutes into the future.
The algorithm was trained on a dataset comprising thousands of hours of films, sports, and shows. It can predict hundreds of actionsrom handshakes to punches. When the algorithm cannot forecast a specific action, it finds a higher-level concept that links them.
Researchers claim this is presently the most accurate method for forecasting events in video. They say the method could bring computers closer to being able to assess situations and make considered decisions rather than act on programmed responses.
Google develops virtual environment for testing video games
Google introduced a machine-learning system for rapid training of game-testing agents, helping engineers uncover serious bugs and allowing testers to focus on more important tasks.
According to company representatives, developers do not need expertise in machine learning. The system works across many popular game genres and can train virtual agents to adapt to the environment in under an hour.
The developers also released an open-source library demonstrating the practical application of these methods.
AI names tennis the most ‘euphoric’ sport
The British platform Freeview conducted a study and found that tennis is the most ‘euphoric’ sport to watch, while football is the most engaging.
Researchers studied emotional responses to various sports categories. They sent material to 150 participants across ten sports, comprising 30,750 frames. While viewers watched clips, a video camera tracked their facial expressions. The reaction was measured using RealEyes’ algorithms.
According to the analysis, tennis is the most spectacular sport to watch at home. Gymnastics, football and cricket follow. Formula 1 elicited the greatest shock among viewers, while boxing inspired disgust.
Heineken developed an autonomous beer fridge that moves to follow its owner
Heineken created a robot B.O.T. that follows the user and carries cold beer.
The autonomous cooler holds up to 12 cans, has six wheels and a camera, through which the AI assesses the surroundings.
The robot will not go on sale, but 20 such devices were raffled among participants in a US promotion.
Boston Dynamics’ robodogs dance to BTS
Seven Spot robots from Boston Dynamics danced to BTS’s hit. Hyundai celebrated the recent closing of the deal to acquire the robot maker.
The choreography proved demanding. A Boston Dynamics employee said the team found it hard to satisfy every requirement, but they choreographed every movement and achieved the desired result.
The robots have demonstrated choreography before. In December 2020, they first danced to the 1960s hit ‘Do You Love Me.’
Also on ForkLog:
- AI revealed hidden connections between the Milky Way and Andromeda.
- WHO formulated six principles for AI in medicine.
- GitHub presented an AI-powered tool for code autocompletion Copilot.
- A study: US federal agencies reckless use facial recognition.
- In Seoul they use AI to identify potential suicides on bridges.
- Scientists developed a model for assessing trust between robots and humans.
Subscribe to ForkLog’s AI news on Telegram: ForkLog AI all AI news!