Artificial Intelligence is introduced to us only through samples of potential hardware, like drones and self-driving cars, that you may be familiar with today. These limitations are due to research about its hardware development remaining in early stages, so its large energy demand did not yet gain solutions.
Presently, AI relies more on software, which energy needs are hard to sustain.
However, Shriram Ramanathan, a Purdue University professor, suggested that incorporating both hardware and software with intelligence features can allow an area of semiconductors like silicon to perform more with a certain amount of energy. It will also balance the needed energy for AI usage in more improvised applications like discovering drugs.
To address the need to control AI’s energy consumption, Ramanathan’s team of engineers developed new hardware capable of acquiring software principles skills.
Human Brain-Inspired “Tree-Like” Software for Artificial Intelligence Hardware
In their published study for the ‘Nature Communications’ journal, Ramanathan and his team exhibited the first artificial “tree-like” memory ever placed at room temperature on potential hardware. It exceeds expectations as previous researches only observed such memory in electronics at low temperatures. This finding indeed puts the hardware closer to the possibility of unloading tasks from the software.
The team developed the hardware using a quantum material, where they instigated a proton called neodymium nickel oxide in the framework as follows:
- They apply an electric impulse to the material, which then causes the mobility of the proton.
- Each orientation of the proton leads to a specific resistance condition.
- Memory states, or information storage sites, are then formed using quantum mechanical effects, where the hardware’s material remains constant as protons position gets rearranged.
- Introducing multiple electric pulses will produce more branches of memory states.
The material’s assets are identified as inexplicable by classical physics. However, they are currently working on the material in their laboratory to learn more about its possible aids in electronics.
Tree-Like Software Inspired by Human Brain Capacity
The software’s “tree-like” description relates to its matter of organizing information into different “branches,” which makes it easy to recover the information when you learn new skills or tasks.
This model gets its inspiration from the way a human brain works at categorizing information to form decisions. Hai-Tian Zhang from Purdue’s College Engineering emphasizes how the brain memorizes concepts using a tree structure of categories. Therefore, replicating this method in the hardware is relevant.
Testing the Material’s Properties and Conclusions
After simulating and testing the material’s properties, the team discovered that the material could learn numbers from zero through nine successfully. This incredible ability to learn numbers, like what the material achieved, is one source test for identifying Artificial Intelligence.
Zhang added that since the material involves protons transport, it can potentially achieve direct, natural interaction with human beings and living organisms. The discovery itself opened up borders brought by the lack of implementation of Artificial Intelligence into electronic hardware. Now that it’s possible, the material opened doors to more opportunities to make a brain implant possible.
In which type of information system do you think you can embed artificial intelligence in a hardware? Let me know in the comments.
Mr. Jaycee De Guzman holds a degree in Computer Science. The machine language is his favorite among the several languages he can fluently speak and write with. As a self-taught computer scientist, he is into computer science, computer engineering, artificial intelligence, game development, space technology, and medical technology. He is also an entrepreneur with businesses in several niches such as, but not limited to, digital marketing, finance, agriculture, and technology.