A global survey of 450 system administrators published today finds 60% of respondents admitting they don’t have enough understanding of artificial intelligence (AI) to practically leverage it.
Conducted by Action1, a provider of a patch management and vulnerability remediation platform, the survey also finds 72% of respondents expressed a need for training but, as of yet, 80% of respondents said their organization is not requiring them to implement AI in their current job role.
Nevertheless, 45% are concerned about becoming obsolete in the job market due to their current level of AI literacy, the survey finds.
Overall, the survey finds AI being used most commonly for log analysis (26%), troubleshooting (25%) and incident detection and remediation (16%), but 64% of respondents that have used AI said they have encountered errors in troubleshooting, followed by 25% of respondents reporting failures in implementing AI for log analysis. A total of 16% of respondents said these failures lead to issues involving prolonged downtime and reduced productivity.
Long term, however, survey respondents identified log analysis, server/CPU/memory monitoring and patch management as the areas they expect AI to have the most impact over the next two years.
Action1 CEO Mike Walters said it’s now only a matter of time before system administrators are automating workflows using AI tools and platforms that previously would have required a DevOps engineer to write and maintain a script. In effect, system administrators will become DevOps engineers, he noted.
In fact, AI will eventually become a utility much like any other that IT professionals will invoke as a service whenever needed to help automate a task, he added.
It’s not clear how IT will be managed in the AI era, but many of the rote tasks that conspire to burnout IT teams will be increasingly automated to a much greater degree. That should substantially reduce the level of burnout that results in higher rates of staffing turnover that many IT leaders currently struggle to reduce.
The challenge will be determining what level of trust to place in generative AI platforms that are generating probabilistic recommendations for tasks that are largely deterministic. In most case, IT teams will need to verify those recommendations are actually applicable, but on the plus side they will not have to build each one from scratch themselves.
Arguably, one of the reasons the management of IT is not already more automated is because building the runbooks required by legacy IT automation platforms simply took too much time and effort. In effect, IT teams are too busy performing manual tasks to find the time needed to automate them. Generative AI tools based on natural language prompts that are easier to master than, for example, a low code tool should make it easier to automate tasks.
In the meantime, however, IT teams should expect to see natural language interfaces being added to every tool they employ. The next challenge will be orchestrating all the AI agents for which those interfaces make it possible to invoke automating a much wider range of complex tasks.
