It's given you some information and now you have to seek out a source to verify that it's correct.
Finding information is hard work. It's why librarian is a valuable skilled profession. What you've done by suggesting that I should "verify" or "proofread" what a glorified, water-wasting Markov chain has given me now entails me looking up that information to verify that it's correct. That's...not quite doubling the work involved but it's adding an unnecessary step.
I could have searched for the source in the first instance. I could have gone to the library and asked for help.
We spent time coming up with a question ("prompt engineering"! hah!), we used up a bunch of electricity for an answer to be generated and now you...want me to search up that answer to find the source? Why did we do the first step?
People got undergraduate degrees - hell, even PhDs - before generative AI.
Look up the tweet from someone who said "Sometimes when coming up with a good prompt for ChatGPT, I sometimes come up with the answer myself without needing to submit".
Verifying information is an order of magnitude easier than compiling it or synthesizing it in the first place. Prompt engineering is an order of magnitude easier still. This is obvious to most people, but apparently it needs to be said.
An entire day of generating responses with ChatGPT uses less water and energy than your morning shower. You seem terribly concerned about signaling the virtues of abstaining from technology use on behalf of purported resource misuse, yet you're sitting at a computer typing away.
You're not a serious person, and you're wasting everyone's time. Please leave the internet and go play with rocks in a cave.
Sometimes you don't need sources to verify something is correct, its something you can directly verify. To reduce it to the easiest version of this, I ask for code to do something, it writes me code, I run my unit test, it passes, my time is saved!
For other things, it depends, but if I'm asking it to do a survey I can look at its results and see if they fit what I'm looking for, check the sources it gives me, etc. People pay analysts/paralegals/assistants to do exactly this kind of work all the time expecting that they will need to check it over. I don't see how this is any different.
I don't think the library/electricity responses are serious but to move on to the point about degrees... people also got those degrees before calculators, before computers, before air travel, before video calls, before the internet, before electricity, yet all of those things assist in creating knowledge. I think its perfectly reasonable to look at these LLMs/chat assistants in the same light: as a tool that can augment human productivity in its own way.
I'm interested to hear more about how you can verify information without a source. What are you looking at when you search for the verification, exactly?