AI had a good run in 2020. Amazing breakthroughs of 2020 made looking forward to 2021 a delight. 2020 showed there are no bounds to technology, particularly AI. The year might be young but development in AI has shown no sign of taking a break. It continues to spawn new developments that will make AI more independent making life easier and better for humans.
Here are Some of the most popular developments so far in 2021
Customized brains for robots
For humans, reacting to our environment is easy. When it suddenly starts raining we know to take cover, we scan the area and look for a shelter close by and hide. Unless we love rain and don’t mind getting wet. Also, interacting with other people is seamless and fluid, we know when a person stretches out their hand as a call for help or a handshake.
For robots, it’s not as easy reacting especially with interacting with people. They might have fast and powerful mechanical motors but their reaction time and response to stimuli is slow
To make robots keep up with humans during interactions, researchers have developed an automated way to design hardware that would increase the speed of a robot’s operation. The system is called robomorphic computing. It assesses the robots physical layout, tasks intended for the robot and suggests an optimized hardware chip architecture that increases the robots speed in responding to stimuli.
Robots calculate everything. They perceive stimuli and try to prepare a response. This takes a lot of computing power and a long time.
The research will be presented by Sabrina Neuman, a recent PhD graduate of MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). According to her, it would be great if there could be a robot that would reduce risk for healthcare workers and patients.
Human brain as AI software, helps computers learn.
Researchers have discovered that the human brain doesn’t need as much information as a computer brain to learn. Computer brains are therefore going to be given more human brain features to learn faster and with little information.
According to Science Daily, Professor of Neuroscience, Maximilian Riesenhube and Dr Joshua Rule have discovered that computer-based intelligence can work like human intelligence when they’re programmed to use faster techniques to learn new objects. According to them, the model would make it possible for artificial neurons to learn new things in a way that could be considered biological. This biologically plausible technique for artificial intelligence to learn will require little examples of visual concepts.
Humans can learn something new by just seeing it. A few months old baby can easily tell the difference between a lion and a tiger with little examples. Maximilian and Joshua are trying to help AI learn this way too. The old learning techniques required a lot of information about a concept for AI to learn that concept.
Super intelligent machines might be beyond the control says, scientists
AI is no doubt a fascinating technology. News of the continuous progress being made in the sphere is truly intriguing but a super-smart artificial machine beyond our control poses danger.
The dangers of AI becoming self-aware and taking over the world is not a new topic. While sci-fi movies do a really good job of telling us what it could look like, researchers are putting facts and figures together to tell us that – super-intelligent machines will be beyond our control.
The study shows that an algorithm that would stop a super-intelligent AI from destroying the world would not work. So according to the calculations, we would not be able to stop the AI from doing that could harm us.
The developments that have been made in AI with just some days into 2021 are laudable. The year already promises to be an eventful one for AI. We might just see more technological advancements indistinguishable from magic.