New technological advances are made every day. This time, researchers found out that machines work better with the use of light. Electricity was the primary tool used to make the AI learn specific commands, but it seems like it’ll become antiquated sooner than expected.
This new discovery has let the researches have access to a whole new world of possibilities. Now, it is possible to improve the speed and functionality of “learning neural networks.” These are a type of artificial intelligence that is intended to copy the human brain’s mechanisms so it can learn tasks by itself.
Currently, all the AI we have access to use a different amount of power depending on the complexity of the task they’re trying to fulfill. The more complex the task, the more power will be used because the data becomes more challenging to process. All these machines have significant limitations due to the slow speed of electronic data transmissions between the processors and the memories.
This research is being continuously done at George Washington University. Through it, researchers were able to discover that the use of photons could be more useful than Electricity to create more efficient and functional AI in the future.
The details of the research were made public in the Applied Psychics Reviews. Some of the juiciest information reveals that the new photon-based TPU (neural network processing units) worked more efficiently than other models. It could perform 2-3 of superior complexity than those of an electric TPU.
According to Mario Miscuglio, one of the minds behind the publication, photonic platforms can achieve the same or better results than tensor processing units can achieve, all of this with the use of an efficient optical memory. However, they will consume more power and will have a higher throughput.
The author continued saying that with the correct training, these platforms will be able to perform interference practically at the light speed. This new application could be used for 5G and 6G networks and include data centers that deal with high amounts of data processing.
This way, photonic processors could be used for saving enormous amounts of energy while at the same time, it will be possible to improve their efficiency.
There have been many recent advances when it comes to AI, and this is one of the most important ones.
Other advancements and discoveries have been made, like the mind-blowing GPT-3. This new tool has been developed to design websites, answer particular questions, and even better prescribe medication. This is a significant advancement, but even if we know that machines will replace humans in individual workplaces, it is implausible they will replace physicians and doctors in the future.
More improvements and new discoveries are made every day so that you can expect more similar news in the following months. Even during times when social distancing is a must, and people struggle to see each other, science doesn’t rest and works on to bring the latest updates every day.