FierceNetwork published a recent op-ed that describes the world’s first autonomous network. Tsinghua University in Bejjing began running its internal network in Level 4 autonomous mode in September. That is an impressive feat. An autonomous network operates without human intervention – it does away with network techs and human oversight. The university network provides service to 16,000 students and staff, making it a network with complexity and varying needs.

This is something that network designers have talked about for decades. Modern networks all have some degree of autonomy and largely run themselves. Any network engineer will tell you that things happen that requires human intervention. The modern network engineer’s job is basically to fix problems that arise and to protect a network from outside interference.

The really big news is that China Mobile recently completed a trial of an autonomous network in Guangdong province, with a population of 122 million. The company has a goal to roll this out nationwide to its billion customers by the end of 2025.

Both Tsinghua University and China Mobile are using Huawei’s Xinghe Intelligent Solution as the software that is operating the autonomous networks. The software uses artificial intelligence to identify risks, resolve problems, and verify changes.

If you think this sounds too much like Skynet in the Terminator movies, the announcement should say that these are really semi-autonomous networks, since the software takes care of about 80% of the problems that arise in the network. That is still an impressive innovation since people are needed to fix most problems in a traditional network. The Huawei solution is trying to tackle every issue that doesn’t require a human to replace hardware or troubleshoot cables.

This announcement will reverberate through carriers in the world, because this level of network automation is huge cost saver by eliminating a lot of the techs and engineers currently needed to support a large network. Obviously, nobody in the West is going to consider the Huawei solution since the U.S. and Europe are boycotting Chinese telecom firms. It’s hard to imagine that labs and vendors here aren’t working towards a similar solution.

There are obvious major risks to  let a large network run itself. There are bound to be software flaws that will inevitably result in a network crash when just the right list of bad things happen. How will a company that has ditched most of its network experts fix an out-of-control network? Network engineers will tell you that they are mostly hired because they learn the nuances of the shortcomings of a network. Is this something that can be duplicated by artificial intelligence – I guess we’ll find out.