AI Accelerator Chips - Not a chip of the old block!!!!
Updated: Mar 24
There was a time when computers took up entire rooms. Many of us still think of computers with Artificial Intelligence (AI) as falling into this category. However, artificial intelligence in many devices (and it is in many devices) requires a chip around the size of a fingernail.
Here, we’ll discuss AI in general as well as AI chips. We’ll discuss what they can do, the devices that they are found in, and what companies are making them.
What Do AI Chips Do?
Perhaps one of the reasons that we may think AI requires large amounts of computing space has to do with a misunderstanding regarding AI. A computer that has AI doesn’t necessarily pass the Turing Test, play chess, or compete on Jeopardy.
What it does do is quickly search through large amounts of information and process that information. Right now, the two main things that AI is used for are cloud computing and computer-to-computer interactions. We’ll cover these ideas in greater detail in the next section.
How AI Chips Are Used and How the Technology Is Growing
AI in cloud computing involves processing large amounts of data from multiple sources. This is used in everything from research to facial recognition to targeting advertising online.
AI in computer-to-computer interactions is made up of two distinct processes. The first is recognizing a given condition. The second is communicating that condition to another device unprompted. This kind of technology is most commonly used in “smart” devices.
These two main uses of AI come together in the oft-cited “Internet of Things” (IoT). This is that most devices will be smart devices communicating with one another on a giant network. Long a popular idea, this concept has received new life with the advent of 5G internet.
Another reason that we tend to overestimate the spatial requirements of AI and underestimate its ubiquity has to do with Moore’s Law. Moore’s basically states that computing power doubles and price goes down by half every two years. This results in computers that are always becoming both exponentially more powerful and exponentially less expensive.
The Most Common Devices that Contain AI Chips.
We’ve already touched down on a number of the devices that commonly include AI chips.
Some of these devices are indeed supercomputers maintained by institutions like IBM. However, they are also increasingly found in smaller applications as well. These commonly include more standard computers and smart devices.
Major AI Chip Manufacturers
Right now, the single largest manufacturer of AI chips is Nvidia. Many encounter this company casually through gaming. However, Nvidia is also heavily involved in the automotive industry including working on self-driving cars – a topic that we will return to toward the end of the article.
The Alphabet Corporation is another well-known manufacturer of AI chips. As the parent company of Google, it makes sense that this organization would have an interest in AI. That’s particularly true since Google offers its own cloud services and its search engine runs off of combing through unimaginably vast amounts of data countless times each day.
Similarly, Microsoft also makes AI chips. While not known for its search engine, Microsoft also has its own cloud service in addition to its line of computers.
Amazon also makes AI chips. As an online retailer, Amazon uses artificial intelligence to do things like suggest purchases and target advertising. It also uses artificial intelligence in its line of smart home assistants.
Other notable chip manufacturers include Intel Corporation and Salesforce.
The Future of AI
Hopefully, this article didn’t leave you disenchanted with AI. Knowing just how simple and ubiquitous some AI technology is shouldn’t take away your respect for this technology.
The Internet of Things has already begun to take hold in the form of smart technology anticipating and preemptively acting on our wants and needs. However, IoT is also the basis for more advanced applications like self-driving cars. After all, self-driving cars will need to know the locations of other cars, obstacles, and even people in order to move without accidents.
Similarly, the basic AI described above is also the basis for more advanced AI applications. Applications like machine learning that may one day facilitate computers making their own decisions on behalf of humans like the poet Richard Brautigan predicted in his 1967 piece “All Watched Over by Machines of Loving Grace.”
To deep dive and stay continuously updated about the most recent global innovations and learn more about applications in your industry, test drive WhatNext now!
Image Courtesy: roboticsandautomationnews.com