Is it safe to write my thesis with ChatGPT? Or will I definitely be exposed?
>>33658427Lost generation
>>33658427No and Yes
there's several different scenarios to what writing a thesis with an llm could entail, and the advice ranges from "yes you'd be stupid not to do that" at one end to "no that's a horrible risk" at the other
>>33658427Ask ChatGPT.
Considering people who didn’t use AI to write their stuff are getting accused of using AI to write it in colleges now. No this is not a good idea. I swear standards have fallen so low that any paper which uses correct grammar and punctuation will get accused of using AI.
Just write it you retard. Use your brain.
Not reading any of the replies, just pointing out that GPT-5 Thinking almost never makes mistakes and that if anyone points out hallucinations, they're an idiot and need to hang themselves immediately.
>>33658427I work as freelance writer and I use ai all the time. Here’s how you use it for academic work:1) It is a research assistant. Use it to help you find sources, grade your thesis, look for weak points. ALWAYS VERIFY ALL SOURCES. Ask it for alternative sources for each one it provides.2) Everything it writes for you MUST BE HIGHLIGHTED IN YELLOW. Have it space its output in sentences per line, then replace each with your own wording (REMOVE YELLOW), then give it several rounds of copy edits. the idea is that go from writer to ensuring they don’t find out. Use an AI detector and see what it flags. rewrite those sections.3) Start with an outline and have it fill in the blanks when you don’t know. I have macros on my streamdeck that copies text, fills in all the xxx’s I leave, and then fills in that text. I ain’t got time for finishing sentences.4) Put in humanizing elements. I’m not saying leaps of logic or typos, but add your own voice. Repeat certain phrases throughout. Ditto if your teacher has seen your previous work. 5) Go to office hours and start passing it off as your own. You may be 3/4 done, but your idea is that you’re not going there to gain any insights; you’re there to play up the human angle. They won’t fail you if you use their ideas or if they have homo leanings. Women know this trick inherently.6) Drop in lefty lingo. Colleges are left-leaning, so use terms that aligns with the teachers interests and activities. It’s not “homeless”, it’s “unhoused”. It’s not poor, it’s “economically disadvantaged”.The truth if the matter is that they used to block students from using the internet, and I’m sure they’re ahitting bricks that their institutions are irrelevant (mostly). I used to work for SimpleTense, a company that writes theses and assignments for Chinese students.
>>33659438gpt-5 regularly and consistently makes mistakes. i use it hours a day. it still gets things like what time it is at the moment wrong, or what day of the week goes along with the numbered day later in the month.
>>33658427Good for ideas, creating a framework, and ordering your thesis, but you can't have ChatGPT write it for you.Many schools have AI checkers and everyone has their own style they write in. If the style of the paper doesn't reflect you, it is pretty obvious.
>>33658427Pretty sure only autists can actually spot AI. I quit my last job upon a new higher level employee sending me five paragraph AI emails that should have been two sentences.Point is, to people who matter, you're probably in the clear. To some random workhorse who gets roped into reading your bs though, they're going to spot it.It's not that hard to just rewrite whatever chatgpt spits out though.
>>33659556Wouldn't it be easier just to write it yourself?
>>33658427Same as anything, it's a tool. If you use it incorrectly, you will fail. If you don't already have an understanding of what it is you're trying to achieve, you won't be able to identify what you're getting from large language models in general. You have to be able to critically analyse the input and the output you're using and getting, and even then you almost certainly can't just ctrl+c ctrl+v the whole thing.AI is best used for inspiration, refinement and as a search engine (NOT a source). It can help you get started, it can help you clean up, but it CANNOT help you if you don't understand what you're doing in the first place. If you couldn't do the work without AI, it won't allow you to do it. If you could do it, AI will help you do it quicker.
>>33661529>Isn’t it easier to—No.Let’s go item by item:1) Research is a bitch and trying to find what you want is worse. That’s like saying walking is the same as driving—isn’t it more dangerous?? Yeah, if you are retarded. And you’ll get lapped by everyone else.2) Getting rid of the yellow is just a placeholder. Unless you’re doing hard science research with empirical data, it’s just jumbling words until they satisfy the gatekeepers. You can prove anything with all of the un/verifiable information out there. 3) Your part in all this is authorship. You’re signing off with your name as this being representative of your work. Just like the walking metaphor above, outsourcing the writing fills in the blanks. As long as the methodology and work is sound, who cares if you used a conjunction not inline with APA/Chicago Style? 4) Fudge it and turn in early versions for review with deliberate mistakes. Make this an oversight, like a numbering order in pages. Then, fix them all after submitting the final documents.5) You’re not going to get graded on how good you are. You’re here to get hired or have recommendations for work. Kiss the ring.6) self-explanatory, though this will change with new mandates.if you just had a bunch of data and you dumped it into an AI, you could say, “Derive several conclusions based on this data and give me 5 ranked theses in order of most plausible.” And before you do that, you also dump all of your old data and papers into an AI project folder and say, “Use my style exclusively and make all new content based on my tone.”NOTE: I believe AI is about to fail big time. Put me in the screencap whatevs, but it’s a recursion problem:I was recently working for a client and I needed a reference for this particular website. So, I searched for sources. The only problem? The author was me (!!!), the article was heavily edited, and it was being used as a reference from work in 2023. The original? LIES!
>>33661534> If you couldn't do the work without AI, it won't allow you to do it.WISHFUL THINKINGYou can do all sorts of things with these new tools. It exponents your output if you’re willing to use it as a learning interface. For example, I needed a quick VB macro for Word. I know jackshit about VB syntax, so I had the code commented out one line at a time and then explained to me. Then, I asked to modify it. And again and again, and it worked. The problem comes when you cut corners, like that Tea app. However, if you just ask it, “hey are there any potential security risk?” It can at least give you a roadmap for redundancy. Hand that off to your IT team and grade their work based in another AI analysis.What I think is happening is that the 3rd-world is using it heavily. And it’s poisoning the well of available info. I’ve found exact translations of my work on Turkish websites.One job I recently applied for is through AI Alignment companies. They are scammy, but I saw what they were trying to do. The only problem is that Filipinos and Indians are monopolizing it with bad English, so things like autocorrect are getting worse. Plus, bad actors know that they can poison the well for non-siloed LLMs, so it starts to be an information warfare landscape. Combine this with all the hallucinations, and I think it will be unusable very soon (without attending to The Camp of the Saints).
>>33661621>I needed a quick VB macro for Word. I know jackshit about VB syntax, so I had the code commented out one line at a time and then explained to me. Then, I asked to modify it. And again and again, and it worked. So you iterated and refined a piece of code, knowing how coding generally works and what you would have to do to test it and what a proper result would look like? Sounds like exactly what I said.But yes, Indians are ruining absolutely everything worldwide, that's definitely true. They earned their hate
It’s a bad idea because if you don’t use your brain it’ll atrophy. Outsourcing to thinking to chatGPT keeps your mind weak. Do you want to be weak? In the future you’ll make worse decisions in life and will be easier to manipulate. It will keep you weak and stupid.
>>33661657> Sounds like exactly what I said.I won’t split hairs. You win.My “hack” for AI/LLMs is to just write my own “infinite macros”, where I can have it prompt almost infinitely by having it deliberately pause (avoid openAI data restrictions), grab text, and continue the process. In this way, I wrote 1,000 SEO articles, so you can extrapolate this data similarly. In fact, I’m sure if I asked ChatGPT to give me a ten-step process to write a thesis/dissertation in a week, it probably could. I wouldn’t rubber stamp it and I think they now manually put in mistakes to slow down users, but you could hire someone in Upwork to sign off on it for less than $500 for 50 pages of double-spaced college slop
>>33658427Im in so much fucking pain i wish i were in coma
You are all retards and I hope you fail in life
>>33661657> But yes, Indians are ruining absolutely everything worldwide, that's definitely true. They earned their hateYeah, the jeet and Chinese menace is on the wane, thankfully. As things get more third-worldly, you can’t just hire Xiao and Rajneesh to fix an already broken system. And with the Grok IP theft, most large companies aren’t playing ball with dual citizens. Walmart just hit sued for Indian hiring practices. Now that the profit motives are stagnant, it’s game over.
>>33661597Not gonna read a post by an AI user. People and researched wrote their own shit for thousands of years up until 2025
>>33661534This
>>33661738You should have used AI to copyedit what your write. Plus, AI and other stuff has been commercially available since 2023ish - you are behind everything.>this train travel is a fad >the only way to get around is horse and buggy
