Student here doing a project on how people in their careers feel about AI — need some help!
Posted by InfamousPerformer100@reddit | sysadmin | View on Reddit | 37 comments
Hey everyone,
So I’m working on a school project and honestly, I’m kinda stuck. I’m supposed to talk to people who are already working, people in their 20s, 30s, 40s, even 60s, about how they feel about learning AI.
Everywhere I look people say “AI this” or “AI that,” but no one really talks about how normal people actually learn it or use it for their jobs. Not just chatbots like how someone in marketing, accounting, or business might use it day-to-day.
The goal is to make a course that helps people in their careers learn AI in a fun, easy way. Something kinda like a game that teaches real skills without being boring. But before I build anything, I need to understand what people actually want to learn or if they even want to learn it at all.
Problem is… I can’t find enough people to talk to.
So I figured I’d try here.
If you’re working right now (or used to), can I ask a few quick questions? Stuff like:
- Do you want to learn how to use AI for your job?
- What would make learning it easier or more fun?
- Or do you just not care about AI at all?
You don’t have to be an expert. I just want honest thoughts. You can drop a comment or DM me if you’d rather keep it private.
Thanks for reading this! I really appreciate anyone who takes a few minutes to help me out.
Kumorigoe@reddit
Sorry, it seems this comment or thread has violated a sub-reddit rule and has been removed by a moderator.
Inappropriate use of, or expectation of the Community.
If you wish to appeal this action please don't hesitate to message the moderation team.
kartmanden@reddit
I use ai for this; Commands, scripts that can be rerun or set up in task manager. I find there is less dependence on one person in my team as I can learn how to do something based on comparing other setup/config/troubleshooting of various types of servers and applications, documentation and input from an LLM I pay around $20 per month for. Lately I have started using Google again as the ai search seems a lot faster. I can do a more broad range of tasks as the information is more easily available. Still, I may need to run a thing or two by a colleague. And sometimes things are so specific and rare that there is no answer but the vendor or a more experienced colleague. Sometimes I’m that more experienced colleague as well.
The fun part: it has made me a lot more curious. Sometimes I could start reading a Wikipedia article then go from there 30-40 steps, diving into more and more details about something. With AI I get a more accurate response, within the constraints I have given. Sometimes when I ask for something and know the topic well and get a poor response, I need to rectify and rerun multiple times.
I hope that people where I’m from understand how it’s really useful. I am slightly sceptical currently about letting an agent loose and doing things unsupervised currently (a system account with access to change something to application/system/user etc), but perhaps down the line.
Traditional-Hall-591@reddit
Fuck AI. My value is in my ability to quickly learn, design, and implement.
Slop generation is a no skill activity, low value activity.
InfamousPerformer100@reddit (OP)
I agree! What is the biggest problem you have with AI? Is it just AI? Or do you think there is something that could be done about it, maybe to fix it?
Traditional-Hall-591@reddit
That it’s called intelligence. It’s a sycophantic prediction engine, basically autocorrect on meth. It had no concept of bad ideas or wrong answers, just the next guess at what the user wants to hear or see. That it is applied to everything. That it amplifies bad ideas and mediocrity.
That it has humans second guessing online conversations. The comment I’m replying to reeks of AI, but humans also use this format of reply.
miuccerundadda@reddit
That’s exactly my stance too. And now myself and other senior devs are fixing slop by juniors who think they’re seniors whom have Ai writing their code.
If developers don’t know how to code, understand the code the Ai has written, they’re not developers and should not be hired anywhere for a development role, period.
No one learns from Ai doing the work for you, when they’re just copy pasting shit it states to do.
Beyond frustrating.
Ai is great, but the way it is being used day to day to for even simple conversational pieces, emails etc or development is wrong. Then you have managers or higher ups pushing the use for Ai because it gets the work done faster. That is false and wrong.
Ai chat bots replacing customer service. No one, especially here in where I am wants to talk to a bot, let alone a phone IVR system. No matter how articulate it is. It’s garbage and a poor experience.
Ai isn’t so much the problem. It’s the lazy people and the retarded managers who enable it and enable the poor use of it.
If you’re a senior who knows what you’re doing. Yeah it’s ok. You can use it to bounce ideas back n forth, have it check your code or implementation of infrastructure or design etc etc as you know what is right and wrong.
But if you’re straight up hey I am working on this complex pipeline or whatever copy paste production code and I need it to do this, how do you it? The persons the problem. Not Ai. Bloody pebcak on another level
Mid 30s here
InfamousPerformer100@reddit (OP)
Super insightful response. I never considered that perspective! Wow...!
"No one learns from Ai doing the work for you, when they’re just copy pasting shit it states to do."
Is there something you do at work or see others doing at work that seem to help with this problem?
miuccerundadda@reddit
I personally do it the old fashioned way and that’s read documentation on how and what something is meant to do or behave so you have an understanding.
That’s how you teach yourself to the point where you do it yourself or at least know what it is needed to complete the job.
And once I’ve actually got a grip on what it is I need to put in place to finish X task, I can bounce it off Ai and reference the documentation that I am reading if I need to.
Test it in a safe environment where it won’t affect other staff or production and where it won’t waste a lot of time.
Plenty-Wonder6092@reddit
Dumb, AI enhances the ability to quickly learn, design, and implement when used properly.
QuantumDiogenes@reddit
I am the IT manager for my company, and I have banned the use of AI across the board. On the broad side of things, this prevents users from exfiltrating PII or NPI, as almost all AI companies have stated that they keep all the data they are fed. On a personal basis, I don't use any AI products as they are useless for the work I do. Every script that LLMs give me is usually wrong somehow. It takes me longer to tell the bot what I want it to do than it takes me to just do it myself.
An example:
If you ask for a script to search an exchange online mailbox, most scripts will return Search-Inbox, which was depreciated a year or so ago. Copilot, to its credit, will return Get-ComplianceSearch, after you tell it Search is deprecated. However, if you ask for a date range search, it will happily return MM/DD/YY format, despite Microsoft Purview only accepting YYYY-MM-DD format. Oh, and Purview will not throw an error, it will simply return zero results.
InfamousPerformer100@reddit (OP)
I have noticed this in my Unix classes as well. AI gives random commands that doesn't even exist. But it does great in simplifying a wordy lab to something I can understand. Is that something that is allowed at your workplace? To let AI simplify or help you understand the problem you are trying to solve? Or is it usually super clear?
QuantumDiogenes@reddit
I have banned AI in my workplace. That means it is not allowed at all. No exceptions.
Calleb_III@reddit
There are no normal people here mate :)
On your question - current iteration of genAI is not AI. It’s a search engine on steroids. And it’s causing irreparable harm to a whole generation’s ability for critical thinking.
Personally I don’t see any task the myself or a colleague can’t do ourselves with better guarantee of it being correct. And don’t really use it nor I intend to use it in its current form.
Call me old school (25 years in the field) or Luddite if you will, history will show who was right amd who was wrong
anikansk@reddit
I think Well Fargo's the best.
MuffinsMcGee124@reddit
I am an application analyst for a Hospital and my only relationship with AI is that our EHR vendor has an AI driven search bot on their knowledge base library. I don’t care much to learn about AI, but will use it to whatever extent is necessary or helpful for my current role or future roles. I am mid 30’s.
MuffinsMcGee124@reddit
And on a more personal level, I hope it fizzles out and stops being so prevalent.
InfamousPerformer100@reddit (OP)
Hey, i want to learn more about that. What about AI do you dislike? I am looking for advice as well because one of my task is to look for something people are struggling with and because i am a student i don't have that insight. Would you be able to provide some. Thanks!
tapemeasured@reddit
Hey, 30s cloud engineer here. In my opinion, the issue isn't exactly with AI. It's with our societal and economic environment.
AI has plenty of issues. There's the recent issue that it's training on AI generated content, creating a feedback loop that degrades it's performance. There's the older issue that it's confidently incorrect, when a human would say something like "I have 80% certainty that we can do this, and I'd recommend approaching it this way", the AI will say "Absolutely we can do this, and here's how to do it".
Knowing when to use AI and how to question it, and perform follow up research, are skills that are developed with a high level of base knowledge. There are limits to AI within a domain, and they are better understood by experienced professionals in a domain.
AI will replace office workers, but only some: the junior and mid level workers. It empowers seniors who can filter out the BS answers, and follow up with their technical knowledge and expertise. However, how do we get senior workers? First, they need to be junior then mid level workers. If there are not enough of those jobs, then in the future there won't be enough senior level workers.
However, that's still 5-10 years away. Corporate higher ups are gambling on the fact that AI will improve enough to replace the senior workers also. Otherwise, as a domain, they won't have enough senior staff to perform the work necessary to keep the businesses running smoothly. Most compensation packages for these decision makers don't factor in these timelines.
There is a need for a junior -> senior pipeline at companies, but that need can't be displayed as a number. Also, junior hires that make it to mid level usually leave to find better paying positions, or different positions. Individuals see this happen, and think "all that work on training that person wasted", then hire a mid level employee to replace them at a higher pay. A new mid level employee that some other company trained up from a junior hire to mid level knowledge.
InfamousPerformer100@reddit (OP)
Yes and that's a real problem! I feel this myself looking for a job lol. AND that's a big bet they are making. I saw a post here that mentioned about AI slop internet theory. AI's just gonna get worse and worse because it's being trained on its own responses. Do higher ups know about this and whats the counterargument to this?
tapemeasured@reddit
Higher ups know what the subject matter experts tell them.
As for the counter argument, now you know why Muffins McGee was hoping it fizzles out and stops being so prevalent.
I don't have any good answers for what we should do. I don't have the time or resources to really look into it. I just see the problems coming up. Companies not wanting to invest in the junior -> senior pipeline is not a new problem, though, AI impact is just another re-hashing of it.
coolest_frog@reddit
IT is a field where we constantly see marketing people come through and promise the next best thing. Before "AI" we had web3 and before that was blockchain and before that it was cloud getting attached to everything. Every time it's pitched as a solution for all problems and it ends up being a good option for certain use cases. Large language models have uses like making emails sounds corporate or speeding up CMD or powershell scripts before we double check them Or meeting transcripts.
InfamousPerformer100@reddit (OP)
This is super helpful! Thank you!
jdoedoe68@reddit
Caveat: I’m far from a sysadmin but I lurk here to learn.
I work in a fast paced 5 person startup. My job is to iterate on proof of concepts and prototypes. I’m having to achieve things I’ve never have to do before.
To me, AI is just a ‘next generation Google’.
When internet search came around, I bet you had people telling you that they just stick to books, or that “I don’t need to google because I know what I need for my job” or “the search results are never what I’m looking for”.
I use AI in three main ways:
(1) When I’m working with something I don’t understand; a new script, a new code repo, or configuration I can ask AI “please explain this to me”. What’s best, is that if I still don’t understand, I can point out my specific misunderstanding, and it’s so much faster than trying to glean the equivalent answers from documentation. Understanding that would have taken me hours to figure out not takes minutes.
(2) As a colleague. I WFH now, but I use to sit with the SREs at work and every hour or so I would ask my colleagues for their input or feedback on my work. Not too dissimilar . It’s not uncommon that I will ask AI: “I am planning to use for my problem. Are there alternatives I should consider?” just as I would ask a colleague. Colleagues aren’t flawless and neither is AI, but there’s always a good suggestion I hadn’t considered in the answers.
(3) To solve code problems. I use Cursor and regularly use AI to build quick visualisations or to debug my code. You can ask it “Hey, I have an edge case where fails. Can you read my code and tell me in what cases thing can occur? “. You can also ask it to build a lot of front end code, to tell you how to get performance improvements from your code, and to even quickly build code for utilities, like “please build me an error handler for ” or “please build me methods for finding the intersections of lines and planes in 3D space”. In these cases, I know what “good looks like” but the AI can spit out faster than I can - especially if I’m working in a language I’m less familiar with.
InfamousPerformer100@reddit (OP)
That analogy with AI and the internet is super helpful. It seems like my problem statement is too broad and needs to be better refined. i have asked this in other threads and get the same responses. This is super helpful thank you. Back to the drawing board!
jdoedoe68@reddit
“AI” as a term means little specific.
Asking how someone uses “AI” for work is a bit like asking someone how they use “fossil fuels” for work.
Of course most people drive a car to work, and they out petrol in their car. They turn on a light switch, and behind the scenes fossil fuels might have been used to generate that electricity.
“AI” is a suite of technologies that change how we interact with existing tools, and in a small set of cases are tools in their own right. That’s it really.
“AI” ( LLMs specifically! ) are very good at allowing you to interact in human language instead of having to click buttons. Let’s say your job consists of clicking the same 10 buttons at the right time 10 times a day. LLMs and agents now allow you to put those 10 button clicks behind a single voice command, or text input. It’s very analogous to having an intern. You can type to a machine “start the computer” just as you might ask that of an intern, and now you don’t have to press those 10 buttons anymore.
For most people, AI is just a new technology added to existing software they use. Use email? Now you can ask AI to write your email for you - but you’re still the one sending it. Use google search? AI will now summarise the results. Do data analysis? “An AI agent can now run 100 different statistical tests against your data and summarise the result so you don’t have to.”
InfamousPerformer100@reddit (OP)
Also thank you for answering the question on how you use AI. I have taken notes on that on Notion.
McDili@reddit
Cybersecurity Engineer at 34 here.
AI is another tool, like anything else it’s to be used in moderation and not depended on.
It more often than not helps me with designers block insofar that I want to design or architect some sort of automation or program. I might give some parameters and requirements and see what it comes up with.
Almost always it comes up with an inefficient design, but then I spot these and now I’m 70% the way there fixing the design and changing it, sometimes overhauling it, based on how unsuitable the output was.
I never supply any overly specific data or any proprietary data. I think some of the n8n flow capabilities out there are pretty cool, but my cybersecurity brain will not go anywhere near it to hook that into any of our data or systems like email for example. Too many exploits with MCP servers in my feeds, it seems like it’s not mature enough yet.
whatever462672@reddit
You need to define your scope better. Deploying faster-whisper for a transcription engine is not the same as asking Copilot to reword an email.
InfamousPerformer100@reddit (OP)
Working on it right now. I am learning a lot from these comments and thank you guys for commenting!
Speed-Tyr@reddit
Yeah sureeeee you are. Def not a bot posting. This post was literally spammed to any technical subreddit at the same time and the account is brand new.
korewarp@reddit
I work as an IT Consultant, and have a technical background.
I'm currently using ChatGippity in my day-to-day, and am looking into Vibe Coding with Claude or Cursor soon.
I'd be down to answer questions.
InfamousPerformer100@reddit (OP)
Do you work with any non-technical background people and what are their feelings about AI and are they interested in learning about it to advance their careers?
yawn1337@reddit
Always the people that just tell others how to work instead of doing actual work
ukulele87@reddit
I think you really need to define your scope, what do you think learning AI means?
Do you think someone in marketing and someone working in IT have the same relationship depth wise with AI?
Are you refering to just using a product that has a LLM behind? Are you refering to creating weights or training our own LLM? Just some "prompt engineering"?
InfamousPerformer100@reddit (OP)
I am trying to identify a problem everyone has with AI and come up with a solution to try and solve for it. It's for one of my school projects. Was going to start off broad and refine from there. Kinda stuck
yawn1337@reddit
I frequently use it to help with scripting. Then I ask to clarify and the entire command changes. Then I just sigh and search up some of the things it said and go from there
trs21219@reddit
I’m a software engineering team lead / devops. In a lead/sr engineers hands it can 10x their output if they know how to manage and prompt it well. Especially if you connect it to your ticket system and assign it work to be done in PRs with very good detail and expectations filled out.
But it’s going to kill the jr/mid market for newcomers.