Google AI's Chip Design Breakthrough

Google is now using AI to design chips faster than humans. That could give it an unfair advantage. Is the chip shortage easing? The world has been grappling with a chip shortage that has hit the production of household electronics, including everything from toasters to washing machines.

If you enjoy articles like these, sign up to the Last Futurist, my speculative blog where we explore AI, stocks, tech innovation and breaking news. Simply put your Email on the top right side of the home page. We are also seeking more columnists so if you like to write pitch me here.

Google's Chip Design AI Breakthrough

Now Google AI has announced a breakthrough in chip design. The AI system outperforms humans in designing floorplans for microchips, according to the Nature paper. Artificial intelligence can help the electronics industry to speed up chip design. But the gains must be shared equitably. However where there is vaccine nationalism, there’s also the weaponization of AI in technology, to achieve a competitive advantage.

The problem is Google’s behavior is giving China and Chinese firms ideas on how to create artificial bottlenecks in the future of technology. Google said in a paper in the journal Nature on Wednesday that a chip that would take humans months to design can be dreamed up by its new AI in less than six hours. But what does that mean for the future of chip manufacturing?

Google is one of the biggest failures in AI ethics, researchers have been leaving Google. So now that a machine-learning system has been trained to place memory blocks in microchip designs the ramifications could be enormous. The system beats human experts at the task and offers the promise of better, more rapidly produced chip designs than are currently possible.

 So far, the floor planning task, in particular, has defied all attempts at automation. That is, until now. It is therefore performed iteratively and painstakingly, over weeks or months, by expert human engineers. Well, say goodbye to those human engineers sooner rather than later. The AI has already been used to develop the latest iteration of Google’s tensor processing unit chips.

• Advancements in AI will be used by companies like Google as a way to achieve a competitive dominance in new markets.
• Think of how they want to make conversational AI the next layer of search. Google search is already considered practically a utility by consumers.

However Google, Amazon and Facebook aren’t regulated properly for abuses in which they favor their own products. Modern chips are a miracle of technology and economics, with billions of transistors laid out and interconnected on a piece of silicon the size of a fingernail. If Google’s AI can do it better than humans, what’s next?

Google’s AI blog boasts about the achievement, of course without talking about any of the repercussions or the AI ethics of having this sort of commercial advantage. It's such a leap forwards it's almost dangerous.

The tech giant’s engineers noted in the paper that the breakthrough could have “major implications” for the semiconductor sector. Nobody is connecting the dots in what these kinds of advances means for society.

One of the many consequences of the COVID-19 pandemic is a global shortage of the microchips that are essential to electronic devices. Bottlenecks actually increase innovation and automation, and the period of digital transformation might actually mean a period of automation of workers is sped up.

The Global Chip Shortage Will Spawn Automation

One well known problem is that microchips are designed in just a handful of companies, including Samsung in South Korea and Intel, NVIDIA and Qualcomm in California. But not all these companies make the chips. 

The biggest manufacturer is the Taiwan Semiconductor Manufacturing Company (TSMC) in Hsinchu, which is responsible for 28% of global capacity using current fabrication methods. Now that they have a pandemic on their hands in Taiwan, what do you suppose this will mean to the global chip shortage?

Ever the more reason for the likes of Google to use AI to do the job in a new and better way. To put it another way, Google is using AI to design chips that can be used to create even more sophisticated AI systems.

Nobody is seeing the dangers of what this will lead to. Optimization of chip placement has been extensively studied for at least six decades. Specifically, Google’s new AI can draw up a chip’s “floorplan.”

• Google is literally laying the groundwork for AI of chips to create better AI.

It takes humans months to optimally design these floorplans but Google’s deep reinforcement learning system — an algorithm that’s trained to take certain actions in order to maximize its chance of earning a reward — can do it with relatively little effort. China, the United States, South Korea and some European countries are increasing investments in microchip research and development. 

It’s too early to know how the shortage will affect the industry in the long term, but the pandemic has focused attention on some key research questions — including how to make the manufacturing process more resilient to shocks and emergencies. If Covid-19 lasts years, so too could the chip shortage. Everything is connected in this world. Goldman Sachs thinks the chip shortage is already getting better.

Google AI Breaks Moore’s Law

The computer industry has famously been driven by Moore’s law — the number of components per chip has roughly doubled every two years. But everything changed on 22 April 2020.

On that day, Mirhoseini et al. posted a preprint of the current paper to the online arXiv repository. It stated that “in under 6 hours, our method can generate placements that are superhuman or comparable” — that is, the method can outperform humans in a startlingly short period of time.

Can you fathom what this could lead to?

• Within days, numerous semiconductor design companies, design tool vendors and academic research groups had launched efforts to understand and replicate the results.
• Mirhoseini and colleagues trained a machine-learning ‘agent’ that can successfully place macro blocks, one by one, into a chip layout. This agent has a brain-inspired architecture known as a deep neural network and is trained using a paradigm called reinforcement learning. 
• Just as AI plays chess in a way that’s baffling to humans, so too do we find AI here exceeding efficiency in seeming unconventional ways. With no more than six extra hours of fine tuning steps, the agent can produce floorplans that are superior to those developed by human experts for existing chips (that look radically different). 

Although the fabrication of the chips is largely automated, the design still relies on manual processes. Google’s researchers have now shown that the process can be completed in less than a day by using artificial intelligence.

• Amazon, Google, Microsoft and other big American technology corporations are using their unregulated monopoly type revenues to pour money into microchip research and AI chip technologies.
• They won’t share them with the world, but profit immensely from this early advantage just as Google will with this new technology.
• The more unregulated our monopolies, the more AI will be used as a bargaining chip in business that creates significant inequality and an AI arms race with China that will lead to significant future geopolitical tensions and dangers.

I’m not certain Mirhoseini understands the ramifications of what he has created.

The AI system was fed 10,000 chip floorplans in order to “learn” what works and what doesn’t. Whereas human chip designers typically lay out components in neat lines, Google’s AI uses a more scattered approach to design its chips.

The reality however is the gains by Google’s AI won’t be shared equitably creating a scenario where China will try to catch up and Western corporations will increasingly “play both sides” in the cold tech war politics of the times. Automation often fuels concerns about a reduction in jobs.

But now we are seeing signs of automation and AI that could displace white collar jobs where significant bottlenecks are occurring. But can we trust Google to use this technology in a good way? That is always the question with monopolies like these, as we have no oversight bodies to make sure they do.