网络和计算机系统管理员

Everyone thinking all systems are going cloud based and companies will not need anyone or anything on site because “AI in cloud” is woefully uninformed on the costs of having everything hosted and ran by those services. They break apart and charge for everything…they’ll do the same for AI the same they’ve done for SaaS, IaaS, etc. At some point the intelligence of AI is going to plateau and backfeed when it has nothing originally to create from anymore. As impressive as it is, I can still see it falling short in many ways.

Current LLMs suffer from substantial unreliability, lack of world models and hence a lack of understanding what something is, how it is related etc. They are simply not trustworthy.
AI-generated coding contains more bugs and errors than human output.
As long as the technology stack behind those models is not radically changed the risk for AI-automation for this job is low, except if a company CEO thinks "quality is not a must". For the IT of this company this would be fatal.

Configuration and security updates of network routers and switches are already being automated by devices like Cisco Catalyst Center.

Due to:
- routine tasks
- a specific set of predefined commands
- Monitoring tasks can be automated
- A business owner's perspective prioritised efficiency and profit over the human factor
- existing example : Juniper Mist AI, Nokia Corteca Cloud etc

Not possible as system hosting will require humans due to Ai not understanding how itself works

so many of the rudimentary maintenance tasks that don't involve having a person there to troubleshoot and resolve will be automated. Eventually, infrastructure will be maintained fully off site and physical robots or AI will handle everything.

A lot of my day to day, is not admin the network or doing patch its landholding users to turn it off and on again, or explaining the same thing for the 47th time to some one who only does that task once a week or a month.

I dont think any of my users have ever used the correct i.t terms to talk about their problems, so i dont see how an A.I could know what there on about, and that when marry says her pc is broke what she really means is she has jammed the printer again.

But I could see more complex tasks to do with patching and licencing etc just going away.

There only thing AI can't replace for quite some time would be physical network infrastructure deployment... By the way, those guys get paid pennies currently for most contracting companies and it is tasked to jr admin most of the time with sr admin having the final review...

As for configurations, routing, switching, network/cyber security, there are no unobtainable variables for an AI that is programmed to do this job specifically. All they would have to do is just get a base of basic job duties and a list of variables to consider when dealing with an A to Z situation and poof, there goes your job. Any unaccounted variables will be added in the testing phases. The real question will be price of AI vs Human and company specific financial situations. Last I check Humans can set their rates lower than what a developer is required to offer it at to prevent net loss.

Servers can't fix themselves

No way too many variables

I took a career in a Management position. If you are afraid to lose your job in the technical field, Management is always a good path. Remember; soon rather than later gentlemen :)

Automation is a crustal part of IT System operations. But configuring new Infrastructure or troubleshooting "broken" networks requires manual Human actions. All repetitive Work is or will be completely automated.

someone got to watch the watchers

The problem of automation here is that the technology changes very fast. Automation can be done in a limited way in an evolving environment.

Creates and manages the communication layer AI uses. Would require either a common available complete drop in network solution that never needs changed or robots to manage the physical side. Neither are less than 20 years away minimum.

Because A.I. and Big Data must rely on Good System / Network Infrastructure either organization will use On-Premise or On-Cloud. However Traditional System Admin can upskills themselves to be Cloud Administrator / Engineer , DevOps Engineer , Security Administrator / Engineer and etc.

I’m a network engineer and agree with the other comment. We have automation and some GUI based systems, but without the underlying knowledge, it’s meaningless. A human will always be needed.

With an AI tranning or eating support tickets the job can be automatized. "Is it working now" can be answered by the customer... The only part of the work that need a human is in case of material problem only...

Lots of work can be automated and implemented via scripts and parts of the code. On the other hand what saves this position is that network topology design changes over time and new implementation methods are needed and all has to be updated.

Systems administrators, or sysadmins, are responsible for deploying, configuring, maintaining, monitoring and managing an organizations servers or other computer systems. Propelled by the DevOps trend, people in these professions have become increasingly reliant on automation in recent years to help them handle mundane tasks like server configuration and software updates. In addition, many monitoring tools are beginning to incorporate machine learning and AI capabilities, which can reduce the number of false alerts and, in some cases, warn sysadmins in advance of problems that are likely to occur so that they can take preventive action. This is likely a position where AI will augment the capabilities of human workers rather than replacing them entirely. The BLS reports median pay for network and computer systems administrators was $79,700 per year in 2016, and the field is experiencing average growth.