I'm curious about how much trust engineers have in AI when it comes to deployment. Modern deployment feels either overly complex with rigid frameworks or too manual, almost making the cloud provider a better option. With AI developing its own alternatives, would you trust an AI tool to deploy an app or manage infrastructure from start to finish? Why or why not? What would need to change for you to feel comfortable relying on AI for deployment? Let's discuss!
5 Answers
"How can you trust a tool with a high failure rate to do something that needs to be flawless?" That's my main concern.
Not a chance. AI struggles with critical thinking and can lead to unreliable results. I'm not a fan of vibe coding for serious applications.
That's fair, but I've had success connecting to my local K8s cluster to resolve issues. AI is decent at diagnosing problems.
Honestly, I wouldn't trust AI with deployment at all. The rate of hallucinations is just too high. I might use AI to help me read and summarize infrastructure details, but letting it actually deploy something? No way!
I get what you mean! But what if AI could create full-stack applications with enough prompting? You still have to understand the process, but coding becomes more conversational.
I might use it to help script deployments, but letting AI decide when to deploy? Absolutely not!
What do you mean by 'deciding when to deploy'? Sounds risky!
I trust AI to write decent Terraform code if it reads the right documentation first. I still check that code myself and would never let it run any commands autonomously. It's a handy tool, but that's where I draw the line.
Exactly! As it stands, my current deployment setup works perfectly fine.