5 answers
7
Questions
Updated
344 views
what are the hard parts of working ai?
is there anything i should be worried of
Login to comment
5 answers
Updated
Liam’s Answer
I think the hardest thing to predict or speak about at this time is to say AI is____ and then it will later be____. The field is so new and we are not sure exactly where it is going to end up even next year. I feel that this is a field where if you skill build you will be good no matter what. If your said goals are to "work in AI" and stay working in AI in one capacity, its not going to be the same later in your career. Cloud computing, HPC, and AI infrastructure are really new fields of computing and in the job market. Be prepared for the entire field to change to something different.
For example. With hardware support, ML hosts were in the data centers but there were only a couple of racks with GPUs in them. A few years later there were hosts with Tesla P4 cards in them that were useful but not popular. In a short amount of time you started to see ARM and co-processing ASICs that were useful to ML workloads, but again they were not scaled. Now there are series of data centers with only GPU containing hosts just to train and host LLMs. This is what I have witnessed over five years. Five years ago there were no data centers that were just GPUs, now there are many. If there is a change like that in five years, where will be in ten years? That's the hardest thing to say, be crafty and stay relevant!
For example. With hardware support, ML hosts were in the data centers but there were only a couple of racks with GPUs in them. A few years later there were hosts with Tesla P4 cards in them that were useful but not popular. In a short amount of time you started to see ARM and co-processing ASICs that were useful to ML workloads, but again they were not scaled. Now there are series of data centers with only GPU containing hosts just to train and host LLMs. This is what I have witnessed over five years. Five years ago there were no data centers that were just GPUs, now there are many. If there is a change like that in five years, where will be in ten years? That's the hardest thing to say, be crafty and stay relevant!
Updated
Mike’s Answer
Pace of the change will be hard to keep up with. But this could be part of the fun as well. Trying to develop a future vision (think 12-18 months, not 5 years) will help you keep pace as you constantly look at "what is next" with this tech.
Updated
Melissa’s Answer
The most complicated part is the uncertainty it creates. It is a useful tool and, in these times, very necessary. However, it’s a bit difficult not to think that at some point it is starting to replace humans in many aspects. We are still indispensable, but the costs we generate are higher compared to what AI can do at a lower cost.
When you learn to work with AI only as a tool, it makes life easier in many ways—whether in the workplace, in school, or in everyday life.
The important thing is to know how to use it correctly and not depend on it 100%.
When you learn to work with AI only as a tool, it makes life easier in many ways—whether in the workplace, in school, or in everyday life.
The important thing is to know how to use it correctly and not depend on it 100%.
Updated
Jennifer’s Answer
AI is a valuable tool, but it is not always 100% accurate. We should always review and verify the information it provides. One challenge is getting the exact results you need, which often requires refining your prompts and providing clear, detailed expectations.
Updated
Shah’s Answer
It's not always completely accurate. We should check the information and its source before sharing what we get from AI. One challenge is creating short and precise prompts to get the information we need.