Is it bad to use Ai for study purposes?

[skeet5]

Hi, is it really that bad to use ai (I use Gemini) for study purposes? I use it to understand a topic, to summarise text I am trying to understand and learn, sometimes to explain the topic I am studying or convert the text it into simpler sentences. I know Ai is bad and I want to stop using it but I just keep coming back to the ai because it helps me sometimes and it is saving me a lot of time. Please can you make me understand why should I or shouldn't use ai for study purposes? Thank you

I don’t have a problem with people using it to rewrite an emotional email or to summarise a long text from someone they cba to read but need to acknowledge to be polite.

but I did say not for studying!
I have had good experiences with the binary questions - like is it legal to do x? Where is x country? What currency do they use in x? How would you translate this word? If you correct and report the ai should learn.

Does asking LLM to redescribe a chunk of text really help you understand it faster and better? * I * don't understand why that would work.

Unless it's legalese, maybe. Askihg LLM to translate legal jargon text into plain English sounds like genius, although you'd also have to tell the prompt to explain all the underlying legal principles too. Might have to that kind of translation in chunks of only 4 sentences at a time.

I'm old fashioned so for me, to rephrase what I heard or read in terms simpler is what helped me learn and understand. Outsourcing that rephrasing to an LLM means I wouldn't be learning. You paid a lot of money in tuition to learn something, then why would you skip on the learning activity?

Thus in lectures when I was at Uni, I would take notes on what I thought I heard & understand, rephrasing into my own words. This allowed me to note things I wanted to ask questions about, too.

I react with "angry" emoji on Facebook each time, so naturally FBk feeds me lots of adverts for AI that will "write your college paper for you". It feels to me that when people use LLMs to "save time" what they mean is they use LLMs to skip a very valuable learning activity they could do and drink coffee and lie in bed all day instead.

He's fooling himself if he thinks it’s improving his learning.

The problem is not the use of AI. The problem is most people are not using it correctly. It’s a tool that is frequently misused, particularly by students who are using it to replace learning.

Schools and universities are teaching students how to use AI for study (not assessments). Go and find some university AI guides, eg https://libguides.bodleian.ox.ac.uk/using-ai-to-support-academic-work/uses-limitations

Yes, it's not good. We are in for a generation or two of fuckwits. It wont be their fault. Try and nip it in the bud and do the work yourself. Trouble is, I expect many teachers are using AI for marking.

Except unfortunately, many, many students are ignoring this and using it to replace learning.

AI is a tool and as such it is fine.
I sometimes use it to generate little quizzes or exam questions, summarise information or something fun like putting certain facts into a silly rhyme to help me remember it.
crucially though, it is one tool amongst many. Don’t rely on it completely.

There are so many issues over this in academia at the moment. I’m a PhD student and use the top section of a google search (AI) when carrying out research and that’s it.

I ignore the AI response to a Google search because on the odd occasion I've read it, it's been full of inaccuracies. (I have a Ph.D.)

I asked a bunch of LLMs a specific health question and asked for links to original sources.

Only one LLM, and only when I insisted on RCT evidence, gave a robust answer. The rest kept saying observational studies or commentaries were RCTs, or insisted on X being true when actually there is no RCT evidence that X is true, and basically quoted speculation as fact.

As a positive, I learned to trust Perplexity more than the other sources.
All the others were scheisse.

No of course it's not bad. It's no different from using Google search. Just don't get it to write your essays!

It's extremely different to using google...

It's no different from using Google search.

AI gives you a lot of words which may or may not be accurate with no idea where they have come from.

Google gives you links and you can judge from the url how accurate or helpful the site is likely to be before you even click on it.

For an academic question, I'd take a url ending in .ac.uk or jstor.org before I'd look at anything on WordPress or Facebook or spewed out by AI. (Some academics write blogs that are worth reading but I probably wouldn't look at the AI at all.)

Similar to wikipedia i use chatgpt, and grok as foundation to use as the very basics then research more credable sources to advance my learning. @skeet5

AI can be an amazing resource for those who are literate, motivated, curious, and research-minded, with the drive to find information through sites or apps that suit their needs. It allows them to explore multiple sources, verify facts, and keep searching if they’re not satisfied, all in the pursuit of what’s true, right, or better. Without these traits, it’s easy to fall short. Traditional methods aren’t necessarily superior either, as they can still be swayed by personal or political biases and may hide or overlook misinformation.

I’ve become such a strong advocate for AI that I often encourage people here to use it for questions or advice instead of relying solely on Mumsnet, especially to avoid negativity and toxicity, while still reminding them to do their own research online.

You do realise that genAI is regularly wrong? And what’s worse is it is confidently wrong so the user is more likely to take its answers as correct.

Having asked it recently to summarise a document I helped to author (a set of accounts) I would currently trust it less far than my 10 year old in terms of understanding, accuracy and general ability to cope. Maybe a different AI engine would have managed better but good grief it was bad (actively inaccurate and cheerfully hallucinating links).

i use chatgpt then combined with grok

But you can't verify facts with AI. It's not like Google where you can look at sources and use your own critical thinking to decide how much you trust them. AI makes stuff up then when you tell it it's wrong, it goes 'Oh sorry yes you're right' and tells you something else. But that might not be true either.

It has its uses but can't be trusted to be accurate.

to me ai is just the starting point the same that people used to say about wikipedia, besides would you trust the newspapers story ? without other sources then why would you trust ai why would you trust wikipedia without other sources ? would you fully trust a tv documentary without other sources ?

and yes most times you can look up what ai is saying because there will be source's if the info exists

It frequently makes up sources.

not the sources ai provides but sources google provides,

not the sources ai provides but sources google provides,

Then go straight to Google and don't waste time on AI.

I wanted to know about a particular place recently. I googled, but out of interest looked at the AI para at the top. It had mixed up two or more places with the same name. I knew enough to know that it was nonsense, but someone else might not.

So basically, you double the work. You ask AI and then you search on Google to find sources so you can check the answers AI gives? That makes no sense.