Andy, many thanks for this video. You just made my ongoing work much easier. In the last days, I tried to chat with ChatGPT, who "kindly" offered me a bunch of useful references that it actually invented (but hey, it apologised for the wrong information, so we're still friends). The tools shown in this video seem to fit much more to what I need. If it will ever happen to meet you, I may need to buy you several pints....
The moment you posted this video, because of your rave review about HeyGPT and you purchasing the lifetime key to HeyGPT Ultra, I actually purchased it too. And I've been using it ever since. But it's disappointing that Chat with PDF frequently fails because the document is too large for the system.
i remember jenni since last summer before chatgpt hype was begun , also scite is really like a producing real scientific articles and proposals easily if you prompt as the necessary questions, especially I wish I had those during my PhD which could have done cut so much time while writing the thesis and spend the best of my times outside before getting old instead of stuck in the lab and office back and forth :/, recent kids are so lucky having these tools freely even for very low cost :)
I still find Bing the best option. I can work directly on a pre-formatted Word document and don't need to copy and paste text between many different tools. Plus it's free and quite convenient when you want to summarise pdfs opened in the browser or extract any information from them. Scispace free plan is also quite good for managing my library and has a gpt chatbot built in so you can chat with any paper and ask questions (copy/paste to your Word document required though). I also wonder how useful Microsoft co-pilot is going to be for research when it's finally released for everyone...
Thank you for another great video! I was wondering if you could cover the ethics of using AI in research in one of your future videos? With the rise of AI, there are important ethical considerations that need to be addressed, such as transparency, accountability, intellectual property, and the line between original text and plagiarism e.g. when it comes to paper writing: Is it still your own text? When does it enter a grey area, and when does it cross the line into plagiarism? I think it's a super interesting and important issue that a lot of people would be into. Thanks again for making awesome videos, and I can't wait to see what's next! :)
1:18 it‘s just because they have very few GPU instances so you‘re being put into a que, leading to lots of cold starts. Common with companies that don‘t have the money for high availability ML pipelines - certainly not „because it is thinking so hard“ 😂
Hey Andy! I have been following your AI cheat codes closely and my last advising meeting went SO well because of it! I was wondering if you could do or if you have done a video on Web of Science. My tiny phd brain isn’t comprehending the website 😅 Edit: Never mind. My library is subscribed and now it all makes sense.
Hello Dr. Stapleton and thank you for all your videos. I am currently a bit confused by how many tools are available and i would like to ask you which one would you chose to start your career as a researcher. Currently i am between Scite and HeyGPT but i am open to any suggestion. Thank you again!
Wow! I don't know if you even saw my recent question about how to include references and citations, but this episode >certainly< answers my question. Well done. Thanks!
Great video. Over all how accurate did you find them when searching for references? I had worked with some tools and I found not very good, or just lifting sentences from papers so in some ways it made it more difficult., and I had them.actually make things up. I do look forward to them getting better. I look forward to future videos
What really bothers me about these systems is that they streamline the process a lot, but they don't really have expert knowledge. From my experience, they only allow for a very shallow engagement with these topics and can not really identify meaningful, new gaps in research. I like them for the exploration of literature because they are more suited than classical databases, but I think for text generation, there is a lack of quality, creativity, and novelty. One tool you missed that I highly regard is asreview.
Whilst I agree, I do think that these tools not being able to identify gaps in research is probably a good thing, as arguably, that is our role as researchers. I think if it did have such power, these tools would quickly become banned in many universities, and we would lose out on the genuine benefits of them. Although I totally agree that they are perfect tools for literature reviews, and the text they do generate is pretty bog-standard. Thank you for the asreview recommendation also, it's great to see something open source still when so many are moving to capitalise on these technologies!
Me: Humans provide quality, creativity, and novelty. AI is just a tool that can help us unleash our potential. Also me: Chatgpt 5 is going to replace us... 😬 😉 P.S. Thank you for asreview recommendation!
Hey Andy, great video. Very helpful. But, I think you have missed SciSpace Copilot in your list of tools that help in research workflows. It comes handy when you want to do extensive research reading and conduct literature reviews. Would be nice to see a walkthrough of that tool in your next video.
Hi, thank you so much for all your content. I'm looking for a Paraphrasing tool that can help me with my writing process in Spanish. Do you have any recommendations? I have my eye on Quillbot, but their Spanish version is not high quality.
Hey Andy, great video. Is it worth it buying the heygpt subscription? Besides your video I could not find any other article or video reviewing this tool.
Greta video. Personally I would advise against using chat GPT, as it is constantly making things up. I provided it a short file and asked for re- formulation. It made up a new text. Elegantly written - but fabricaton. Beware of that.
Interesting and highly informative video. Do you think there's still a potential for startups to spawn out of gpt to service academia. I've been thinking of starting something similar to these websites you are reviewing. One idea with disruptive potential is to use gpt like technology for materials mining. You probably know what I'm talking about.
Some offer a free (limited) option but the processing power required to run AI can be quite expensive (I think I read ChatGPT is running at over $100K per day) - I think we'll see more affordable option as time goes on.
Can you use Scite alone to write papers? Or is it just for references? I've been seeking something to find & Adding citations throughout the paper. That's probably the most tedious task
Thank you for the amazing content you provide i hope you will have the time to reply to my comment as my deadline is thr 15th of july. I have finished writing the manuscript however i have some few questions : 1. Which Ai tools that you suggest that i should start with to proofread and highlight my mistake and suggest modification if needed among all since my manuscript is 3000 words almost and most prompt of Ai do not accept more than 1000 words - 2000 words i think except bing Ai in creatibe mode. 2. What is the best tools that translate or if i wanted the Ai to create an abstract out of my manuscript what would be the best you think? Since i want ti translate it to french PLEASE reply as soon as you can.
But sometimes with ChatGPT finding academic references, you then ask it "are you sure?", and it apologises for getting it wrong. Then it tells you the opposite findings were the case. I don't trust it. 😐
Try some "tree of thought" prompts using a scenario of three experts to help fact check the information. Also if I say this is wrong at usually fact-checks itself again.
I can see now that anyone has started to create AI tools through API to earn money.. I can see dark age of academia from now on... We'll witness enormous number of junk and fake pseudoscientific papers :) which look nicely written but shallow in content.. originality and creativity will degrade gradually..
Argueably we already reached that, with pressures to publish there have been so many hollow, and cookie cutter papers coming out in so many journals. While journals make a profit, and universities put pressure on researchers, this will be the case. AI didn't cause this, but I agree that it will contribute exacerbating it, mostly by helping workers meet increasingly unhinged publishing KPIs
I think the opposite will (also) happen. The ability to leverage AI to do prodigious academic work is going to explode how much GOOD research we'll see.
with the paid version you can access chatgpt 4 which can read more information, produce more writing AND importantly access the internet - version 4 is most likely to give you accurate references. Whether it's worth the money depends on your finances, and how you use it
A major problem with AI is that it recycles existing knowledge and does not create new knowledge. In the end, only PhD's may be safe from the onslaught of AI.
AI developer here: the most dangerous would be precisely the day AI will be able to generate new fresh knowledge, since it would then be called singularity. There, you could write that a major problem happened
Great! What about Ai jobloss? Or ai as weapons? Ai job agents everywhere. Shouldn’t we cease Ai / GPT immediately? Or at least start Pausing Ai before it’s too late?
Lovely, bunch of BS anyways last 20-30 years, all powered by money, most research is falsified, science is old, we need to rid it. Time for the era of Artificial Intelligence yay!
Didnt prepare for this one huh? If you want someone to set aside their att6ention to watch you, please prepare better. This was not your best work bro.
I am now writing my master thesis and I have ChatGPT4 in my back. Tbh I am quite nervous the get some kind of plagiarism since it generates much better text then I do.
@@isakhammer6558 My simple advise, use it for cracking the complex subject you don't understand, then sort out the literature and from there you can synthesize the information and present it in our own tongue. Hahaaah.
@@anthonygikuri wait u guys aren’t allowed to see your own similarity report? Mine let us see the similarity percentage whenever we hand in a version of the essay and we are free to reupload it as many times as we want after modifying it (though after the first three times the report will not instantly generate and have to wait for a day) Or is the chatgpt similarity not included in the report itself