By Ravi Kumar
This Modlingua Live talk compares the fastest supercomputers of the time with the human brain. On YouTube the video is titled "Supercomputers Through the Ages: The Race for Unprecedented Speed"; the original page title asked who wins the supremacy war between supercomputers and the human brain.
A short history of fast machines
The speaker begins with Japan's K computer, announced in 2011 as the first machine to pass 10 petaflops, or 10 quadrillion calculations per second. He then describes how China's Tianhe-2 and Sunway TaihuLight raised the bar, and how IBM's Summit, built for the US Department of Energy, took the lead again in 2018.
He also mentions supercomputing work in other countries, including NASA systems in the United States, Jugene in Germany and the Cineca consortium in Italy. The figures in the talk, such as Summit's speed and size, reflect the time it was recorded. Supercomputer rankings change often, so check a current list before quoting them.
What supercomputers are used for
According to the talk, these machines support climate research, disaster prevention, medicine, physics and astrophysics, and the search for new materials. The speaker's point is that they let scientists work with data on a scale that people could not handle alone.
The human brain as a comparison
The second half turns to the brain. The speaker gives rough estimates, including about 86 billion neurons, and suggests that the brain's overall processing power is still beyond the best machines. These are popular estimates for comparison, not exact measurements, and experts disagree about how to compare brains with computers.
His conclusion is practical: people should value their own abilities and use them to solve difficult problems for the benefit of others. He ends by saying the channel is preparing a further video on artificial intelligence and human values.
Questions to take away
- What tasks suit a supercomputer, and which need human judgement?
- Why is it hard to compare a brain with a computer using one number?
- How can languages and translation benefit from fast computing?
Related reading
For the speaker's related talk on machines and human values, read Humanoids, Humans and the Future of Education.
