A.I. vs humans: who can design chips better?


“Perhaps we should all stop for a moment and focus not only on making our AI better and more successful but also on the benefit of humanity”

by Stephen Hawking at Web Summit in Lisbon, November 2017

To come up with a productive, compact, and most importantly efficient computer chip, requires a lot of hard work and effort:

  1. First, learn the basics,
  2. Then gain years of experience through trial and error,
  3. And in addition, keep an eye on new products on the market in order to overtake competitors.

Well, or just use AI.

In June 2021, the journal Nature published an article where Google Brain researchers introduced a deep reinforcement learning technique for floorplanning, the process of arranging the placement of different components of computer chips.

This essentially involves plotting where components like CPUs, GPUs, and memory are placed on the silicon die in relation to one another — their positioning on these miniscule boards is important as it affects the chip’s power consumption and processing speed. 10,000 chip floorplans were fed to the AI system in order to “learn” what works and what doesn’t.

This new method automatically generates chip plans in less than 6 hours. The resulting chips are superior or comparable to those made by humans in all key metrics: from indicators of performance and energy consumption to design and area. Besides saving time, AI chips can help engineers experimenting with tweaks to the code along with different circuit layouts to find the optimal configuration of both.

The AI has already been used to develop the next iteration of Google’s tensor processing unit chips, which are used to run AI-related tasks, Google said.  

There are other big companies testing AI development tools that help place components and wiring on complex chips.

For example, Nvidia started making graphics cards for gamers but quickly saw the potential of the same chips for running powerful machine-learning algorithms, and it is now a leading maker of high-end AI chips.

We believe that there is always room for improvement, and we are interested in applying new methods of working with AI in the future. After all, it is a way to improve efficiency and optimize production.

But are the capabilities of the human brain have been exhausted?

Of course, it is not the case. 

AI hardware and software will continue to require abstract thinking, finding the right problems to solve, and choosing the right kind of data to validate the solutions. This is the field where better AI chips cannot outsmart humans. 

Engineers still need considerable experience to control, since amplification algorithms can behave in unpredictable ways. Without human supervision, this is the path to costly design or manufacturing errors. For example, game-playing reinforcement learning algorithms can fixate on a strategy that leads to short-term gains but fails in a long term.

The main thing is the synergy between human and artificial intelligence. If there’s a right way, it’s one of AI and humans cooperating. So, let’s create awesome projects together! Our years of experience and competent use of new technologies will open up opportunities for your equipment.