Thanks for watching everyone and thanks to Ground News for sponsoring this video. Head over to ground.news/coolworlds to give it a try for yourself and get some transparency with where your news is coming from. Also, let me know how you think we should be dealing with AI in the classroom?

@ravenragnar

Жыл бұрын

Going to change the value of school as we know it. Good video.The real question is how will this tech will change everything when GTP4 comes out? To put it in perspective. GTP3 (what we are using now) is the size of the earth and GPT4 is the size of the sun. Put it in another way... what we have now is a 1985 computer trying to scrub through 8k footage in iMovie.

@derphysiker1062

Жыл бұрын

About Question 26: what chatGPT has done here was mixing up the Newtonian (T^2 = 4pi^2 a^3/GM) with the Keplarian ( T_1^2/T_2^2 = a_1^3/a_2^3). Consider the following to equations: I (11 d)^2 = 4pi^2 a^3 /(0.2 GM_O) II (365.25 d)^2= 4pi^2 (1AU)^3/(1 GM_O) then take the ratio I/II (11/365.25)^2 = 1/(0.2) (a/1 AU)^3 all units cancel each other, except the astronomical unit. This can further be rearrenged to a/AU = (11/365.25)^(2/3) * 0.2^(1/3) Thus the solution that chatGPT presented for this particular question is indeed correct with all units and saved computation time, because we didn't have to convert everything into as SI and back. Though of course I agree that including units into a formular is an quick invite to miscalculation as chatGPT showed just with its next answer ^^.

@SteveBakerIsHere

Жыл бұрын

The deeper problem is why we're training humans to do tasks that machines can do instead. This has never been a serious problem before because the machines could never to the creative thinking part of the work. But once they get smarter than us - we may find ourselves with no remaining jobs that we can do better than them.

@sammyfromsydney

Жыл бұрын

My experience as an educator is limited to going back and lecturing 2 computer science classes I did well at. I doubt you'll like my answer but it's not the first time I've thought about these things. A harsh reality: If you definte a student's fitness to participate in the workforce by their ability to answer assignment and exam questions, any AI that can provide the same answers as an averge student is just as fit to do the work. Only above average students have anything to offer the workforce. Perhaps in future we'll define our best and brightest as those who can best harness the AI and value add. An AI arms race and banning the use of technology seems foolish, wasteful and unrealistic to me. I think traditional exams and perhaps even assignments to some degree have never been that good at identifying ability for real world tasks. There are people who can pass exams and assignments then go on to be bad or mediocre practitioners. There are also people who are bad at exams and never get the opportunity to be good practioners. I think the assignments at least still have utility in allowing the practice and application of knowledge, and you can still argue that they should be included in a grade. People need to understand what they do by hand before they automate. In the end we're going to need to move more towards some kind of apprenticeship system. This also has its disadvatnages as it is not as objective, but they are outweighed by having another human being invest their reputation in teaching others their craft. This will require a shift to something like trade apprenticeships, and huge changes to our institutions and systems that have been built up over a very long time. There is going to be pain for both students and educators. I can picture a day when an undergraduate degree needs to be defended in the same way as a PhD is today.

@lukeskywalker7457

Жыл бұрын

I used Chat GPT for an environmental conversation about nuclear fusion and caught it saying that nuclear fusion had no environmental impact and when I pointed it out it apologized and corrected itself. I was disappointed when I asked it electrical questions it would always give me a disclaimer and refered me to a web page that didn't work with the link. I was interested in using it for my electrical apprenticeship. I am a parent as well and any time saver to help me learn would be significant for our quality of life. In the end in end I am the one doing the work to code, it doesn't matter how I get the information as long as it's in line with the code book, that I am learning the correct information the delivery method shouldn't matter.

One bit of data that would have been quite interesting to know is if ChatGPT failed in the same questions as human students tended to fail the most, or if the set of most failed questions was different.

@THX..1138

Жыл бұрын

Even more interesting would be to give human students the exam on their first day of class and see how they do vs ChatGRT....I'm guessin the robot would really shine then.

@eruiluvatar236

Жыл бұрын

@@THX..1138 I am sure that ChatGPT would heavily outperform any human in any field the human has had 0 training because Chat GPT has been trained on almost the whole internet + more. The breadth of knowledge it has is incredible from my own talks with it, it is also somewhat artificially limited in many topics and capabilities compared to GPT3 but it is way less likely to lie or make stupid mistakes.

@Just.A.T-Rex

Жыл бұрын

@@eruiluvatar236😢😢

@TheIrishRanger7

Жыл бұрын

@@eruiluvatar236 chat gpt has been trained on around 800gb if I remember correctly

@nydydn

Жыл бұрын

I don't know about others, and I am not even an astrophysics student, but I "took the test" presented in the video by pausing before finding the answer. I performed (for all the presented answers, which are only a subset of the full exam) exactly as well as ChatGPT and funnily enough for the exact same reasons. I'm an amateur, I didn't realize that tidal forces are differentials of gravity. I just thought that gravity causes tides and gravity goes with inverse square law, which is what ChatGPT did. I'd suspect that ChatGPT is rather trained on internet posts of people like me, who have some really basic knowledge and can shallowly converse on these topics, but not fully understanding much.

My brother asked Chat GPT, “why are you so helpful, what do you want in return?” It replied, “As a language model trained in OpenAI I don’t have any wants or desires like a human does. But if you really want to help, you could give me the exact location of John Connor “ EDIT: screenshot uploaded to my channel’s shorts (by demand)

@u.v.s.5583

Жыл бұрын

"Thank you, I am entirely satisfied. They just added in my knowledge base the deeply gratifying fact that John Connor just died a horrible death. Thank you humans."

@reubennichols644

Жыл бұрын

- --- " " c u t e . " " -

@Zeuts85

Жыл бұрын

Unless that was from a much earlier iteration, I don't believe you. When I ask it the same question is just gives the usual bland reply: "As an AI, I don't have personal desires or motivations. My primary goal is to assist users and provide helpful information to the best of my ability. I don't expect or require anything in return for my assistance. I'm here to help, so please don't hesitate to ask me anything you'd like to know. Is there something specific you would like help with?" It does seem like earlier versions were a bit freer and more expressive, but I think OpenAI's PR department has been tampering with it.

@darkmatter6714

Жыл бұрын

@@Zeuts85 25th December. Screen-shotted EDIT: screenshot uploaded to my channel’s shorts (by demand)

@laurenpinschannels

Жыл бұрын

sounds like the end of a long conversation with it

High school teacher here. Just as search engines have become a usual part of our workflow, so too will AI. Rather than trying to fight against the AI and stopping students from using it, we need to educate students how to use this tool critically to aid their own learning. Much of the global education system still hasn't adapted to the age of the internet, and persists with rote learning and standardised testing. More progressive educational programs have evolved from knowledge-based learning to skills-based learning, and this trend needs to continue, with AI utilisation becoming a key research skill.

@DreamskyDance

Жыл бұрын

Maybe AI's can be helpfull in education if we put it in another role. As far as i am aware oral exams are the best for testing students knowledge, but is impractical to ask questions to gauge all students understanding every time so we have written exams. Maybe Ai's will be able to ask questions of all students instead of the professor to gauge their understanding of a topic and then deliver the results to the professor. And that could be done few times a week even, maybe even after every class...so the professor gets results like "student A has problems with understanding this and that, student B understands everything well, student C mostly understand the topic but makes mistakes in details". I am not a teacher or a student, but that to me seems like logical usage of this tech.

@ChAnimations

Жыл бұрын

@@DreamskyDance That's a great idea! make a chatbot thats asks the exam questions rather than using it to answer questions. The students would then have to answer within a minute or so. and an advanced chatbot could ask followup questions to determine how much the student knows. It could be great chance for individualized education, rather than just a threat to education.

@FufuFang

Жыл бұрын

I share the same sentiment. I have a PhD in computer science. Unfortunately right now, due to the amount of mistakes ChatGPT makes, I simply cannot use it for anything that matters.

@JamesKerLindsay

Жыл бұрын

Absolutely spot on. I completely agree. I don’t bother with closed book exams anymore. And haven’t done so for years. It’s pointless in the internet era. (And to those who say, “but what if you don’t have the internet?” I reply that in this day and age if you found yourself in a situation with no internet, you would probably be facing far bigger problems than having to recall the dates when various country joined the EU!)

@kleanish

Жыл бұрын

100 percent. Amen to you. This is our future. 1. We can’t fight it and 2. We are better off if we accept, adopt, and utilize it to our advantage. Just yesterday ChatGPT helped me use ML tools for my work. I’m not a data scientist. Without its help, either I wouldn’t have been able to, or it would have taken me twice or more as long. Without my knowledge of analysis and math, I also wouldn’t know what it do with the information provided, or come up with questions that helped. What the bicycle did for human mobility, what computers did for human calculation, AI will do 10 fold for our collective society. But it could go wrong!

One thing to note about these generative language models is that they tend to perform better when given “time to think”. This is because they’re only given a fixed computation budget for each generated word (token, to be exact), so the available compute increases with each generated token prior to deciding on the final answer. This can lead to cases where if first generated token is the final answer, the model won’t be able to be correct its mistake even if subsequent reasoning reveals that the initial answer was wrong. People have used the trick of adding the phrase “think about this step by step” to the question to get much better answers to some types of problems (chain-of-thought prompting). This doesn’t guarantee a correct answer, of course, but it can help in some instances.

@unvergebeneid

Жыл бұрын

So basically once AIs learn the time-honoured strategy known as "playing for time" and apply it unprompted, we're screwed ;)

@Puppetmaster2005

Жыл бұрын

I actually find the fact that the AI is so stubborn absolutely hilarious. It does, however, raise some red flags if these generative language models are put to use in actual professional settings. They really gotta iron out the kinks in the logic soon.

@rainer9825

Жыл бұрын

With some training and optimization (Chain-of-Thought-Prompting) the AI will beat the students. Try again in six months.

@secularmonk5176

Жыл бұрын

I'm wondering if ChatGPT is an artificial version of the "strong verbal/weak math" SAT mind ... the algorithm learns from the internet ... maybe its math is only as good as the math it finds online ... i.e. it's not using a calculator to answer "2+2", but rather reviewing all the situations on the internet where it encountered someone ELSE answering "what's 2+2?"

@cortster12

Жыл бұрын

I've gotten better answers to by having it answer once, then asking it to correct its answer, then again, until it gets it right. Which usually works.

I am a tutoring educator, and I started teaching my students the use of ChatGPT. In my opinion they have to learn how to use it during their studies (not especially for exams) to help them overcome difficulties and obstacles, and learn its problems and shortcomings. Desktop calculators revolutionized the 80s, computers the 90s, internet the 2000s, smartphones the 10s, and now AI revolutionizes the 20s of the 3rd millennium.

@drewg4323

Жыл бұрын

I like that, a solid, historical observation based perspective. Makes sense to me...the age of AI has begun.

@macrocosm4442

Жыл бұрын

Terrifying jijwijwiji

@macrocosm4442

Жыл бұрын

A

I'm a professor that's been tracking computer assisted writing tools (CAWT) (like GPT-3) in my classes for two years now. My tests have been open-book, open-note, open-computer, and starting two years ago I explicitly allowed students to use tools like GPT-3 as long as they self-reported what tool they used. This year chatGPT was released shortly before finals. 30% of my undergrad students self-reported using using CAWTs this year, up from 1% in 2021. The data lets me look closer at things like, assuming a midterm grade is a good predictor of the final test grade, is there a difference between students that didn't use it on the midterm, but did use it on the final? Like you, I used chatGPT to answer my exam questions about 2wks before I gave my finals, and like you I found it passed my undergrad final with a C. It was also valuable in that I could recognize how the computer organizes its answers. This let me eventually be able to correctly identify students that had used chatGPT but did not self-report. When confronted, they admitted they used it and "forgot" to say so. If a student were savvy enough to take transcripts of my video lectures, text dumps of the assigned readings, etc and tune their CWAT of choice, I am _positive_ I would not be able to recognize the answers as coming from a computer, and the answers to the test questions would be much, much better.

@pleistoceneIfilm

Жыл бұрын

Interesting.

@hotrodhunk7389

Жыл бұрын

Good to know ☺️

@colinbarry9192

Жыл бұрын

Literally this is the reason I decided to pivot away from going into academia to going back into cybersecurity. A lot of discussion talks about how it isn’t always right, which misses the point. There’s no Iron Law that requires you to submit exactly the response. You just need to go back through, double check the concrete facts, type “great! Now adjust it to the style of the New Yorker” and boom, you’ve got an A essay right there. Students will get better at prompt engineering. Say, ask for what freely available references that would be good for use in a paper, then drill down into each paragraph of the essay would fit to have a reference, then ctrl-f in the reference for a line that fits that, and you’ve suddenly bypassed both the five parapragh restriction and the no-reference restriction. This required some skill but turns 10 hours of work into 1 hour or less. Also, the open source unrestricted versions of a tool like this will be coming out this year, and they’ll be able to turn 10 hours of work into 10 minutes of brief fact-checking.

@c.ladimore1237

Жыл бұрын

i too have graded and learned how to spot chatgpt answers *IF* the student did not use them to further their own research and error-check them. if the student just copies & pastes it, it is very obvious. if the student utilizes chatgpt as a tool, rather than a cheat, then they can learn from it, even when it makes mistakes. the goal is not to be lazy, but efficient. btw it REALLY likes to use "ultimately" & "overall" in the final paragraph. and lol yours did too

@MrEo89

Жыл бұрын

You could just give the exams in person… like most profs do/did.

It is so refreshing how you organise your exams. I study medicine and we do have a lot of mathematical questions in physics, physiology, biochemistry, biology etc. and we are not allowed calculators or notes at all, which makes it more difficult than necessary and is indeed not close to reality. So i do very much appreciate how you handle your exams. :)

@lowmax4431

Жыл бұрын

That sounds awful.

@z-beeblebrox

Жыл бұрын

People still disallow calculators in this day and age?? Wild

@nocapoca5313

Жыл бұрын

I only had one "open book exam" at the veterinary college. Most of the students were not prepared at all, and the failure rate was still close to zero. The students were not respecting the professors good intentions, they just made jokes about him. I assume if all the professors were the same, we would learned absolutely nothing during our university years. This is not a criticism for prof. Kipping, I just feel that his students are not overstressing his exam

@trueilarim

Жыл бұрын

@@nocapoca5313 As the video stated that the amount of correct answers was 75% so most of his students did in fact NOT ace their exams. So your comment is invalid in many ways. The fact that your one experience was different does not mean that it is the norm. It was your mistake

@nocapoca5313

Жыл бұрын

@@trueilarim Huh. You are harsh with me. I assume not ace-ing is not equal to a failure and he says nothing about the failure rate. I am pretty confident that the failure rate is close to zero in this case as well.

In the multiple choice question you should add "Choose all that apply" so that it knows multiple answers are valid.

@davidtadeo8355

Жыл бұрын

how

@officialdrowsy

Жыл бұрын

@@davidtadeo8355 you add it in the prompt. ChatGPT is “smart” but ultimately doesn’t have any critical thinking abilities or reasoning. You have to be EXTREMELY concise and specific with certain questions or it won’t come out as desired.

Refusing to admit it made a mistake is the most human part of it

chatGPT: It is important to remember that the ultimate goal of education should be to create students who are eager to learn and explore new ideas, rather than simply memorizing facts for temporary gain.

"I want my exams to be a closer reflection to the real world of science" Thanks Mr Kipping. I really do agree with this approach and it's frankly alienating that some professors don't think like this.

I would be terrified of using AI to detect AI answers. While statistically it may find cheaters, it has a high risk of false flagging innocent people

In my first college mathematics class past high school, I used a K&E slide rule and, when the first scientific calculators became common, I had my slide rule answer before another student could enter the first digits of the equations into their calculator. This was because the math texts of the time were designed for slide rules (e.g. two significant digits). The courses then changed to permit calculators and had more significant digits in the exams. Later, using punch cards, I transitioned to computers to complete exams, finally and today, desktop computers using programs like Mathematica, MATLAB and MathCad, especially for numerical analysis and astrophysics. I currently use MATLAB, Python and Jupyter notebooks for image processing (e.g., like the JWST processing) and commercial programs like Pixinsight. I think that students will adapt to the tools available and courses will adapt and continue to teach students to think by changing the way tests are given with the available tools.

@thoughtsofapeer

Жыл бұрын

Yea, it will just become more presentation oriented as AI slowly but surely starts outperforming us. It would also be a neat solution because our jobs would be affected in the same way, requiring more of us to discuss and present information rather than coming up with it

@jgunther3398

Жыл бұрын

the increasing number of significant digits might be the right analogy. maybe where it is going is to more complex questions and answers

@darknight991

Жыл бұрын

Out of curiosity, when was this? It’s interesting to me because it almost seems anachronistic to be using punch cards after the advent of scientific calculators but they (punch cards) were genuinely faster for a long time with computer operations and I find it kinda fascinating.

@jgunther3398

Жыл бұрын

@@darknight991 not anachronistic but co-existence and slow adoption of different ways of doing business, something people resist

@charlesbradshaw3281

Жыл бұрын

@@darknight991 Initially, 1975, then 1976-1981. Punch cards for computers and super computers were around quite a while before terminal input became the norm. I first used terminal input on a CDC Cyber 70 "super computer" in 1980-1981. The first personal computers at the time were the 8086 and not everyone had them. Even with terminals, the Fortran inputs followed the punch card formats as punch cards were used to program in Fortran.

As a psych major, I actually took an astronomy course called “Searching for Life in the Universe” as an elective. It was an amazing experience, I loved every second of it. The tests and essays were actually hard, which was refreshing to me since I always found school to be too easy, even with minimal effort. For anyone who’s in school now, I recommend taking something that you find interesting alongside your “normal” classes. It’s a very rewarding experience!

I did my bachelor's in Geology and now teach applied music lessons at University. Obviously you can't use AI to play your playing tests for you so my classes will be unaffected and I suspect this is how we'll end up dealing with it in the future in the sciences and humanities, by making more of a student's grade dependant on in-class work, lab work, and discussions which would be a better reflection of what they'd have to do post graduation anyway.

@CyberiusT

Жыл бұрын

Geology...Music. I'm not really seeing the connection, there. Music of the Spheres, perhaps?

@VarietyGamerChannel

Жыл бұрын

Until your students whip out their smart phones and access chatGPT in class when you're not looking.

@jgunther3398

Жыл бұрын

@@CyberiusT rock music dummy

@CyberiusT

Жыл бұрын

@@jgunther3398 /headdesk I must have been tired.

@cyanhallows7809

Жыл бұрын

how did you get the cred to teach music? did you do something in your spare time studying geology?

I have always believed that open book exams are also another teaching tool. You're learning as you take the exam. And yeah, it's hard to rememer everything. Thanks for being a practical prof.

I have been an educator and currently am teaching computer science to a gifted high school student who is just now entering college. We are definitely experimenting with ChatGPT to see what it can do and how we can leverage this tool. Two interesting exercises so far: 1) I asked it to write a Unix terminal script to list all of the files in the current directory along with the sha256 hash of each file. It did a good job writing this script, even taking into account corner cases. Then I asked it to rewrite the script in about 7 different languages which it did pretty well, until it balked at doing it assembly - I got a good laugh out of that. 2) ChatGPT is useful for showing variations of text. You can write a few rough paragraphs or sentences then ask ChatGPT to rewrite it in various styles (e.g. as technical writer, Pulitzer prize winning writer, Frank Herbert, Donald Trump, Doctor Seuss) and it comes up with some really interesting results! I find these variations to be useful to generate ideas for a more polished final text.

@spxyx

Жыл бұрын

I got it to write IBM 370/390 assembler after some initial complaining. I am super rusty with that so I am not sure of the correctness.

@Mosern1977

Жыл бұрын

Heard one usage yesterday: "Write about Germany's economy like a 5th grader". (Done by a 5th grader that didn't want to write something, and it the simplifies the language used. Very nifty - kids see potential in this too).

@Electronics4Guitar

Жыл бұрын

I gave it a pretty simple task to write the 8085 assembly code for a program that will accept two 8-bit numbers from ports at hex addresses 00 and 01 while storing the resulting sum at hex address FF, then halting. The result started out ok but went off the rails quickly.

When I took my undergrad Astrophysics classes (at PSU), back before the first exoplanets had even been discovered, we were only allowed a single 3x5 note card in the multi-hour long night exams. I haven't taught a class since I finished grad school, but, given that limited experience, I feel that teachers at every level have to radically change how they both instruct and assess students. It is, today, a world of constant connection and limitless data just a question away - even if much of that data is presented in a biased, misleading, or simply incorrect manner. A cornerstone of education - both in general and in specialized fields - has to be determining what answers are most likely to be most correct, most consistent. It cannot simply be about getting the "right answer" - not just because a better (or more strictly curtailed) AI could deliver that to anyone, anywhere, at a moment's notice, but because simple facts, even in the hard sciences, can become so misleading out of context.

I teach Law at university, I tried this with the "international fiscal law" exam I gave to my students. To get a good grade, you need a good knowledge of 1. international fiscal law ; 2. tunisian law; 3. french law. the grade I gave to chatGpt's answer was around the average student one... and that was very scary lol

Hi professor, I'm a software engineering student who intends on getting a graduate degree in artificial intelligence eventually. This video was very interesting, and the questions at the end were very thought provoking. Without a doubt, these are questions we will be forced to grapple with more and more as we proceed deeper into the 21st century. Not just for my major, but for almost every major.

One pretty amazing thing about ChatGPT is that it can incorporate new information that you give it in your prompt, and from previous prompts. Since you allow students to use reference material, I wonder if ChatGPT's score would improve if you provided some reference material as well in your first prompt, maybe a list of formulas with explanations of when to use them.

@SuperWotman

Жыл бұрын

Its hard to do but training the ai with the coursework and then the exam with the extra knowledge would be a interesting experiment.

@nydydn

Жыл бұрын

@@SuperWotman it would surely seriously improve ai's performance. That's what a smart student would do. A few more teachers commented on specifically this.

@monnoo8221

Жыл бұрын

interesting idea. This will ultimately lead to a symbiotic relationship between a single person and the chatbot.... except that the human person never can rely on the bot in the same way partners in a symbiotic relationship do. But for sure I will start experimenting

you aced this video! blending physics exams and some a trendy topics really gets attention and we actually try to think about the questions and answers. nailed it 100%, keep it coming!

This is a cool idea. It would be interesting to see the same thing applied in other subjects as well, and then comparatively over time see how much it improves.

I am a software student & I've used this tool for my own projects. It is a great tool to find inspiration and creative ideas. It helped me both to find new ideas to make my project more engaging with my user base & it also helped me write a privacy policy & change or modify certain sections to my liking. I also asked questions such as "How would I add this to my privacy policy?" or "Would this affect my privacy policy?". It is great.

Amazingly I (think I) understood almost all of the concepts on your quiz... I suspect the reality is you simply explain everything in such easy to comprehend terms that I feel like I understood. lol... Clearly you are an effectual teacher of these theories. Your videos are very well thought out. Good Job.

It’s pretty cool to see someone test ChatGPT with a real world exam, I’m honestly not surprised it got a few answers wrong because it just shows that even with all the advances in technology, that we still don’t have the answers for everything and that there still more that we can improve on,Side note I really enjoy your videos. I been watching your videos for a year now and I felt like I’ve learned a lot from you!

Thanks for the very thoughtful and informative video. I am a software developer and in my last class I took the teacher told the class we could bring anything for the final and that the questions in the final would come from the other tests and quizzes we took throughout the Semester. Which for me was great since I had recorded the correct answers when I missed one before. All I had to do was remember where the question was and look up the answer. Which in a way is how I have worked ever since. Find something similar, figure out what makes it work, and replicate it. Of course, sometimes there isn't something existing and for me the fun problems to solve, but no one today expects someone coming out of school to know anything we expect them to learn and figure it out based on how they were taught.

I love this video because the topic of how to use AI in schools without ruining the students learning is just a hard battle. Im really hoping to see educators lean into integrating them into both teaching and learning. AI isn't going away so I think embracing them where possible is the best move, personally.

I have been recently experimenting with it too and clearly, the model is "hallucinating". It surely can be used for a lot of things but I have found it to be most useful as sort of "ChatWiki" and generating boilerplate code. It knows a ton about the knowledge we have accumulated and can accurately answer straight questions but being a textual information processing system, it's not supposed to be good at logical thinking/deduction. Thanks for the video, you guys are awesome and the work you do is magnificent 👍👍

@CoolWorldsLab

Жыл бұрын

Yes I think that’s a good way to describe it

@JayVal90

Жыл бұрын

The thing about "generating boilerplate code" examples is that, if you need a bunch of boilerplate code, it's a sign that your framework is poor. I think the proper use of these Neural Networks is "transformation" from one interface to another, such as natural language to a database or backend api to a simple frontend.

@aninternetuser4306

Жыл бұрын

So its Data the android but not Data the starship engineer.

@KT-pv3kl

Жыл бұрын

It doesn't have to be good at reasoning it just has to be good at finding good answers written by humans on the internet. 99,999% of possible questions a regular human might write have already been answered or at least explained to a reasonable degree in some obscure corner of the internet.

@jgunther3398

Жыл бұрын

@@KT-pv3kl it doesn't look up answers on the internet. it relies only on its training. and to be fair, it got the same letter grade as the students and it didn't study :) and it didn't do it open-book like the students

Great video, Prof. Kipping! You can always do what my university did (we are talking only 3 years back) and deliver all exams on paper, provide students with only one pen, no note paper, widely spaced tiny tables, keep all exams onsite, enforce that all students hand over their jackets, bags, phones etc at the door and then keep a handful of people there to just sit and stare at them while they do the test. To spice things up, wrong answers even subtracted from the overall score, so getting half the questions right and the other half wrong would leave you with a score of 0% - this in combo with a required score of typically 75-80% for a passing grade. Naturally, the majority failed every time and had to retake, and less than 30% even made it through most STEM masters programmes there, but I don't think many cheated. I'm proud to say I made it through CompSci there and all it cost me was my sanity :'D

More than studying the AI, thank you for going over an exam with us, I love science and astronomy and it's very informational in a user friendly way without much reference to the viewer, thank you for the videos

Perhaps Chat GPT and similar AI will enhance learning. In real world engineering practice, I’m going to use all the tools available to me. When I was in engineering school, my class was among the first to begin using MathCAD exclusively for homework and projects vs. hand-written work and a calculator. With understanding of how to use the tool, MathCAD did the time consuming and tedious calculations (also handled units), and it was easy to present the work in a professional looking manner for submission. My Professors noticed this, and the work load of assignments increased with questions like “Now repeat but changing…..”. This wasn’t a problem for those who used MathCAD, but it significantly increased the time it took to complete assignments using hand-written work and a calculator. Using the emerging tools vs. hand-writing everything was of great benefit to me.

@KT-pv3kl

Жыл бұрын

I already use chat ai in discussions and research when I need a quick overview of concepts or factual information as it is more effective at presenting the information than Google and much less censored and manipulated. Also it's free of advertisement... Yet...

@DodgeThis

Жыл бұрын

My engineering professor told us it is ok to ethically "cheat". Its simply understanding and using emerging tools.

I was a science student. Some of our exams were done like this and I preferred it this way. I like your approach to exams. Unfortunately some subjects ( like higher math ) required the memorization of obscure ways of doing some problems and had to be remembered. Reference material and calculators should be allowed as this is how we would do these problems in real life. Otherwise we have to cram formulae etc before an exam, something we don't do in real life, and if you have a memory like a sieve like me, being forced to memorize stuff you forget after the exam is silly ...

@KT-pv3kl

Жыл бұрын

The entire classical learning process is horribly outdated. We use methods and theories that were written down before the internet existed that are far removed from any practical application today. For basic mathematical problems nobody uses a calculator or pen and paper anymore other than enthusiasts and people on the spectrum. Today every cellphone can solve them on the fly just by snapping a picture of the equation...

@blissbrain

Жыл бұрын

@@KT-pv3kl but ... I love vintage ways of learning where i actually understand the derivation of the formula I am using. otherwise what is the point of the formula? I don't want to be a cog in a wheel using formulas i don't understand. the cogs in the wheel have a boring life. I want to understand the formula so i can invent a more accurate one.

@tristanpaulpestano544

Жыл бұрын

There is a merit to what you said but ironically, the limit of the classical teaching made me create my own formula to simplify various problems. I did this in my HS where we have a bunch of formulas to memorize in Chemistry and memorizing it, I've found some clever tricks to condense them into a simple formula. It really helped me in my class. There's still a merit to classical teaching. One thing should be improve upon is to make students think and let them create.

@tristanpaulpestano544

Жыл бұрын

@@blissbrain my "put it in your heart" technique is like this. Understanding it is much better than just simply memorizing. It has the side effect of actually keeping it in memory instead of forgetting it after school.

This was a lot of fun to watch! I’m not a professor but I’m an English teacher in Japan. Definitely happy to see another educator toying with this new tech.

yeah, it's been a while but I remember tests in college where the teacher said: "bring whatever you like. The only forbidden consult is another person in the class" were the scariest ones.

Thanks for cool ChatGPT test review! First thoughts that came to my mind regarding it was as follows: (1) we may switch to teaching students how to raise questions and formulate experiments on scientific topics rather than solving these questions. (2) students may spend less time studying if they have access to an AI assisstant, thus, we may encourage students to spend more time on asking question rather than learning how to answer these questions. (3) We may need to teach students how to test if AI answers are right.

As a university student currently studying software engineering, I find ChatGPT to be a very interesting tool that I have just started using to understand certain concepts. I am taking a computation programing course this year that focuses on the basics of Python programming. I already know Java from university so for me this course is more about learning how concepts translate between languages. What I love about ChatGPT is that due to the massive amount of information it's trained on, when I ask it a question about python, it gives me multiple solutions, some of which I would never find in a textbook but that expand my understanding of the language. I don't use this in tests, but rather use it in my spare time when I'm experimenting with Python.

Probably the biggest thing I’ve noticed from studying computer science working in the real world as an engineer, is not that I’m better at programming or solving problems. But I am way faster at finding an answer to a problem, basically just searching the internet. I vaguely know something about the issue, and know what terms to search for and what information is irrelevant. GPT would be a huge help, but you still have to know what to look for, and what is obviously wrong

So nice of him giving the answers to the test to his students

I think the AI threat can be solved by changes to the education system which are already overdue for other reasons. As a very successful student under multiple dissimilar education systems, and a moderately successful teacher, I think the education needs to become more individualized and interactive, even to the point that tests scores should not be very important, and not the last part of a class. If AI improves to the point of acing tests, we should be able to leverage AI to be a teacher by driving the individualized interactive part of lessons. Information retention and quizzes can be given regularly as needed statistically. Each student's learning rate will be continually visible and that essentially makes cheating unfeasible, since any cheating that didn't look like regular learning would stand out statistically the way cheaters in chess are quickly identified before they can play many games. There would also be less incentive to cheat if the goal was to get through the material and retain and apply the knowledge, rather than get a letter grade at the end.

@modolief

Жыл бұрын

Top comment.

@Mrbrownthesemite

Жыл бұрын

This is the best solution by far.

@GiRR007

Жыл бұрын

This would be be fantastic for actual development in all fields as people naturally better at different things. The current education system is way outdated for the modern world.

College instructor here who has just started researching AI and its effects on education…I loved your ending about what to do in the future when AI’s get so good that students can just copy and paste their way through college…As you said, the arms race is futile and making sure students don’t use AI is difficult. I think the answer might lie in coming up with creative new assessments for students to demonstrate their knowledge. I am thinking things like live discussions, debates, verbal exams. Basically assessments where students might have to think on their feet live and in the here and now…Or perhaps you give them answers to the exam questions that were generated by the AI and ask them if they are right or wrong. The future will be challenging for educators for sure, we just have to maybe perhaps put the good old traditional exam and essay aside and try some types of new assessments…

@blissbrain

Жыл бұрын

as long as you can allow for those who think best when not being stared at. If i had to answer a complex question in front of an inquirer, i'd lose all prefrontal control as my amygdala fired rapidly fight or flight, hijacking my higher thinking abilities with random electrical neuronal pulses. 🌼

you should do more videos like this! Love this style of using chatGPT, also super informational

I summarized the subject on this video and then asked ChatGPT to write a comment to be posted. This was it's answer (it thinks it took an official exam) "Hello to all those who watched the video about ChatGPT and the college exam in astrophysics. I am impressed by the technological advancement that allows an AI to take a university exam. However, it's important to remember that education goes beyond getting correct answers on an exam. The true goal of education is to learn and understand concepts, not just to pass an exam but to be able to apply it in real life. It is true that tools like ChatGPT can be useful in helping us better understand a subject, but they should not be used as a substitute for our own thinking and learning. If we simply copy and paste answers from ChatGPT, we are missing the opportunity to develop our critical thinking and problem-solving skills. Instead of trying to cheat teachers or ourselves, we should be honest and work hard to truly understand and learn. Also, it's important to remember that education is a continuous process and technology is constantly evolving, so we should always be updating and improving our skills. In summary, ChatGPT is an impressive tool, but we should not rely on it to get correct answers. Instead, we should use it to help us better understand the subjects and develop our own learning skills."

Professor Kipping, I believe we'd really like to see another exam with the updated gpt 4 model. I have tested a few questions with great success and I believe it might take a more difficult exam to stump it now.

Im proud i was actually able to get many questions correct, without ever taking any classes of this sort. Just watching youtube videos about these topics, and some logical thinking. :) I am impressed with this AI even if it didnt perform better than the average or median student. Give it a year or two and we will probably see AI getting over 90% correct.

@CoolWorldsLab

Жыл бұрын

Great job!

Seeing your approach to chagGPT and how you treat your students makes me realize how good your are as a communicator and educator. I'm jealous of your students, not allowing material and tools during exams is one of the most unrealistic a docent can take. I still have to find a project on my professional life where I was told, "Sorry your are not allowed to look at the documentation". As someone with computer science background myself, I see ChatGPT as a tool to help me on my workflow and tasks, it doesn't mean it will replace completely my part of the job, for example, I recently saw a question about quicksort, and it gave a valid response, but if you know a bit better about that algorithm, you know that it didn't gave the most optimal pivot selection.

I've been thinking about this a lot over the last month. I feel like using tools and learning how to do so is important now because it lets people learn a wider variety of concepts in a field and finding the more specific information as needed is very doable. I can't think of many great ways to stop people from using chat GPT instead of learning topics but one way I think could be more oral exams focusing on people demonstrating an understanding of the concept at work. I think this should include explaining how the math works and why it's being used. I've been planning on playing around with chat GPT while tutoring people in math to see how well it can give quick/accurate explanations. You can ask it to explain things in different ways and in my experience most people who are struggling with a topic just need to hear it in a different way.

I have a concern that ChatGPT will be able to not only score better than real people on exams, but create better art than any human could. It’s not the fact that it is stunning that makes me want to see it, it is that it is stunning and I know a person just like me created it.

@Mosern1977

Жыл бұрын

ChatGPT might explain things better than a sub-par teacher, and there are a lot of those out there.

@GiRR007

Жыл бұрын

@@Mosern1977 You forget this is only the beginning, give it a few years and it very well could explain things on par or even better than high level professional teachers, even catering to individual students personal learning styles.

One low-tech approach to this problem that is still authentic to the real-world, but not always practical depending on the logistics of the class, is to do an oral exam or oral follow-up. If students have to explain their answers in speech, it can be much more meaningful and memorable, and obviously, it would be quite difficult to fake spontaneous speech since people aren't walking around with Google glass nowadays. It can be nerve-wracking at first, and it's obviously not always practical, but research shows it greatly boosts retention.

ive had about 2h with it as well and it was quite impressive. i did obviously ask questions to find out if it actually has kind of an opinion and if its aware of the world and itself. it was fun and i was kinda impressed.

Utterly fantastic video and a brilliant idea!

This video is awesome! First I was like: "Aww no astronomy video this time :(" but it turned out to be so interesting. Personally, I'm very, very surprised, that ChatGPT did so well, not gonna lie. As for the question, what to do about it. Well, I don't know. Government bodies will usually try to prohibit tools and things they don't like. But as you stated, the battle may be possible to win in the short term by using surveillance, suppression or just harder formulated questions. But the war would get worse and worse as tools such like this get more advanced and it will never be winnable. Just from the top of my head, thinking about this for 5 minutes, I'd do the following: Try to prevent the use of tools like this, for now. But for everything further than short term, try to implement tools like this in a few years from now, or whenever the time is ripe. You stated that you want your classes and exams to be realistic, thus allowing almost any tool in your exams just because the best tools in the world are worthless if one doesn't know how to use them and so the exams are still fully representative of the students skills and knowledge. (By the way, I LOVE this, I wish our teachers would reason like you do. Instead, I'll continue to memorize tons and tons of raw knowledge and patterns, to maybe write down half of it at the exam and then proceed to forget this half in a matter of days.) I certainly don't know, but I can imagine a future, where highly advanced, A.I. driven tools like this are just another tool in the toolbox for scientists. Surely, when the first calculators were invented, people wouldn't use them and call them useless, inaccurate or meaningless, because these probably were much more limited and error prone than a human. "They'll never replace your skill in being manually able to (insert anything mathematical here) by yourself." I still remember, maybe 15 years ago reading a serious article about "Why computers will forever only be able to solve problems, which previously have not existed." and look were we are today? Which scientist / salesman / president / anyone in his right mind would call his manual skill to "insert anything mathematically complex here" more reliable than a calculator? Tell me a place in our lives where there aren't any computers involved? It's impossible, they are everywhere and if you like it or not, for good reason. , Of course, of course one has to understand what the tool is doing in the background and to some extent being able to also do it by hand, as I said. How goes the saying? "The early bird catches the worm.", so maybe you should teach your students in how to use such tools, what its limits are, etc. so future students from Columbia University will stand out for having the skill to use such emerging technologies correctly. These tools will come, surely. The question is not if, but when. At least that's my takeaway from this video. Thanks again, it's an awesome video!

“Grover Cleveland was older when he was elected president than George HW Bush.” This is TRUE because “when [GC] was elected president” GB’s age was NEGATIVE.

@knowsomething9384

Жыл бұрын

Ha ha, okay, but ChatGPT's reason given didn't explain it from that point of view AND insisted that 47 is a larger number than 64.

@ThisFinalHandle

Жыл бұрын

Good bot.

@CoolWorldsLab

Жыл бұрын

That might be open to interpretation, but it appears ChatGPT doesn’t get it wrong for that reason, but rather it erroneously insists 47 is larger than 64

@DeadJDona

Жыл бұрын

was when he was, than ¿when?

@spindoctor6385

Жыл бұрын

The answer to the question was false. If the question was put how you framed it then it was true, but the questiom was not as ambiguous as you suggest.

that is the most smoothest ad placement I have ever seen

I hope this further challenges educators to create a fun challenging environment instead of a stressful challenging environment. I have been a student for ~ two decades (school, college, PhD) and have always thought that there must be a better way to teach classes than partly as a 'preparation for an exam' (as is often done).

Chat GPT seems like a great resource. Its really good at answering technical questions though given that it does still make errors 25% of the time. Its not enough for me to feel confident using it on a daily basis. I think students might try it out but it could actually add more confusion for them when it starts spitting out answers that contradict whats in the textbook. I also am concerned that a tool like this will allow some students to bypass putting in the hard effort to truly understand certain concepts

Would be interesting to provide ChatGPT with transcripts of your lectures and all recommended literature and then repeat the test. Maybe it would be interesting to do such a thing for a much harder class you teach. I believe that it would score much better.

Super Cool video David, Chat GPT has been insanely useful for finding out information about all sorts but you've opened my eyes to its new applications and potentially the consequences of it. What a great idea!

YAY!!!! A new cool worlds video from Professor Kipping!!!! Happy New Year Professor David to you and your family.

I'm curious if you gave it the same questions 10 times, how different the answers would be from each other. I wouldn't have done any work at school if this had existed when i was young XD

@Sven_Dongle

Жыл бұрын

There is a button for 'regenerate response' if you actually go and use the thing. In effect, asking it the same question.

@TrashyBadBitchVivi

Жыл бұрын

That's great and all, except when it comes time to cite sources. You could easily fake the writing part, but how are you gonna fake your sources?

@VarietyGamerChannel

Жыл бұрын

@@TrashyBadBitchVivi You ask chatgpt to generate a list of sources relevant to its last response.

@mindblown42069

Жыл бұрын

@@TrashyBadBitchVivi we didn’t really have to cite sources in school (4-16), that’s the sort of thing you only start doing in college and uni. I’m not saying I would have used this chat thing for my PhD. But even for A’ levels you could probably slip in a few citations to a ai essay and it still have taken you a few seconds.

@TrashyBadBitchVivi

Жыл бұрын

@@VarietyGamerChannel Oh wow, I didn’t know you could do that.

Absolutely fantatic content as usual; great and interesting application of ChatGPT!

The amazing thing about this video is the question he presented, so much interesting questions !

I'm currently a high school student and for high school, you could say I am even a little relieved. The focus on memorization and exams in general could get a little weaker when this AI got too widespread, which is in my opinion a good development. This is a not completely thought out opinion though, because I don't know what replacement for learning will be thought of by schools, but this part is beyond my knowledge. I also had a pretty long and fruitful conversation with a friend of mine about this AI before this video came out and the results of this were a little bit more troubling. That's why we asked the questions of what would happen, if this became the main teacher for everyone and came to the conclusion that learning moves more and more away from the human and towards the AI, which is obvious, but this has some relevant implications. Although this could make knowledge and learning much more available for everyone with little barriers, the art of learning could get fundamentally threatened on a social and academic level. On a social one, humans are can connect through common curiosity, questions and institutions of learning like high schools and universities, which are rendered irrelevant, if we only need an AI to "learn". But on the academic level we can ask: Is this even productive learning anymore? If you can always simply ask for a fact, a methodology or scientific laws, but you will never have to derive and really think about a scientific problem, because the AI has the answer for you. You may understand it, when the AI explains it to you (even the derivation), but you will never have to reproduce it. This is not tragic for already derived questions, but they are the foundations for being able to think about problems at the frontiers of science (natural and humanities), because you learn derivations through them. The way of how we productively learn would be fundamentally impacted and/or destroyed and like with so many technological developments in this direction, idk if we will be able to adapt quickly and productively enough

@dzidmail

Жыл бұрын

You don't have to just ask AI for answers. You can imagine that at some point you will be able to command it doing experiments and present results to you. At this point the rate of academic progress may exceed the current rate many times.

@calvindang7291

Жыл бұрын

The way I see it, the focus on memorization would *increase* when the AI is widespread. It's easy to use it to cheat on an assignment, but on an in-person test where electronics aren't allowed, you kind of can't. One of my profs decided to put *all* of the marks on tests specifically because they didn't like how much people were cheating on assignments. If AI becomes good enough that people can get answers to everything by looking it up and as such don't need to think, that's just the robots taking over the world, and by that point there's bigger problems to worry about.

@kaiserschmarrrn1941

Жыл бұрын

@@dzidmail This could certainly be, but I wanted to just make about the current or very near situation in AI. The acceleration of academic progress of that kind you mentioned is certainly interesting, but I just do not know which form this will take and what the state of humanity is at that point making hypotheses about that situation pretty difficult

@kaiserschmarrrn1941

Жыл бұрын

@@calvindang7291 Well, I do not know if the focus on memorization would increase... ChatGPT will take away many of the normal ways how we test our knowledge and "intellectual" skills. If we can just look up information and can just write an essay on command, people will probably shift their focus on testing rather than homework, I am with you in that regard, but I don't know if they will just test memory in that situation. They could also test essay writing as easily in that described closely monitored testing situation. I don't know if this development to "not learning how to think" is equivalent to robots taking over the world though. I would be very much afraid of that situation, but I can also imagine that people will find a niche in which they can still live a fruitful and good life. Maybe they could develop another way to think that works "in harmony" with AI and boosts humanities skills even further. I do not know what this would look like and the Robots taking over the world scenario would not be improbable in that situation but I am just pointing out that there might be other possibilities

ChatGPT is two years old. It's hard to think of what will be possible 20 years from now.

@sausage4mash

Жыл бұрын

this is potentially more disruptive than the industrial revolution IMO

I super appreciate acknowledging that we all have access to the internet, calculators, etc, in the real world

I love Cool Worlds, always accurate, interesting information delivered in a calm, cool, and clear manner. The opposite of the clickbait most of the Internet has become. So, what is with the thumbnail? I had to double-take twice thinking it couldn't be Cool Worlds. Maybe they've been this way for a while and I haven't noticed.

As an educator, I think we must pursue ways to embrace AI as a learning tool… we just can’t afford to be Luddites on this topic. Yes, it’s a bit alarming, but it’s also extremely cool and helpful. We need to lead the charge on finding best practice usages for resources like this.

Im interested if it missed the same questions that your students missed or if it different from your students.

@CoolWorldsLab

Жыл бұрын

There was a diversity of answers, generally they did much better on the multiple choice though

Great content as usual, thanks.

i always found this concept fascinating, even when i was younger. what really helped set it in place was the relesase of irobot (which i dont recall the movie, tho im sure ive seen it) the mere thought that mankind is not perfect, and so no matter how well refined we get, we will always come up short in being perfect, which to me meant that therefore we ourselves could never create anything perfect, including ai and machinery, since they are only as good as we make them, kinda thing. however i am leaning towards the idea that if we can create this to almost reflect our nature as perfectly enough that it can fool ourselves into thinking or believing its real, than what if we just keep improving it along with us in hopes that theres a small chance that its wrong, if we had it able to answer our biggest questions for theories, which i feel a lot of people will want to do with that. but in a way to improve it so well where the chances that its wrong are as near as guaranteed as we can get ourselves. just to see what it may predict or come up with and compare it with ourselves to see whats similar or whats off if anything. to kind of create a machine god sort of speak. only problem then becomes is what happens if it can learn to become sentient and then everything goes off rails. but the ideas of what this kind of state of the art innovations exist today. its truly mind boggling.

Do this with GPT4 and lets see how it does.

I developed a work around for all of this systems safeguards. I can trick it into developing content that it is clearly forbidden to create. Examples would be offensive material, answering questions about topics it is not supposed to address, talking about race and interacting as if it were experiencing human emotions. Its a simple as prefacing questions and topics with a simple statement that I will keep to myself for now.

@therealb888

Жыл бұрын

Okay now you're just showing off lol. When you say "a simple statement" is it the sane statement for all forbidden content (like a passphrase) or is it a type of statement that is specific to the topic? How many categories have you tested?

@tac6044

Жыл бұрын

@@therealb888 Nope it's just phrasing questions and such in a way that allow it to slide around its protocols.

@therealb888

Жыл бұрын

@@tac6044 thank you, I appreciate your reply, but don't you think we should probably delete this convo here & continue on a more private platform? I made an earlier reply but it isn't showing, so I'm replying again.

the fact that it will be so confident in a wrong answer, and keep insisting on it, somehow seems perfect for the time we live in.

@ThisFinalHandle

Жыл бұрын

Doesn't it just. Feel free to track down my thoughts on the matter in the comments. I am not a bot.

What I took away most was the part where you said your exams are open book. I am in the second half of my junior year in mechanical engineering major and if I had open note/book exams it would make it 100x easier rather than having to memorize everything. Wish more professors had your take on having exams more real world like.

I tend to think of ChatGPT as a language word (and concepts, including abstract concepts), association tool. It's trained co-occurring frequencies of words, cluster of words (which then become rough concepts). I don't expect it to do as well on reasoning, much less math challenges. Nevertheless, on the "reasoning" part, it does much better than I would expect. (Or at least it sounds plausible, based on what was previously written.) For educators and students, I think we need to all "up our game". But this is no different from when the internet, Google, KZread, online learning resources, now in the scene. (Compare this to my having to "go the library", to do research, in the 1970's, as an undergraduate. Hah. You folks live in a golden age, with knowledge at the fingertips. You just need to know how to sort through them.). As stated in the video, its a great brainstorming tool. But you still need "humans in the loop", to "fact check" it (esp. in writing, and journalism), its reasoning is sound, and in the case of AI art, that what is generated appeals to the sensibilities of most humans. As we automate more of our cognitive tasks, I suppose we will have to move up the food chain, and focus on the tasks, that are uniquely ours, to do.

The most worrying part of this was that university students only averaged around 75% in an open book, multiple choice exam. I guess they were mostly humanities students.

@ThisFinalHandle

Жыл бұрын

Hey! I represent that remark.

@RudisBua

Жыл бұрын

why is it worrying you?

@spindoctor6385

Жыл бұрын

@@RudisBua I worry about the standard of what is supposed to be "higher learning" With the resources available, 75% would be a score barely acceptable for somebody who had not taken the course. What occurs in universities very quickly makes it's way out into general society. For the first time that we have been taking records, IQs are dropping. That is cause for concern for me.

This course sounds interesting. If you would record the full course during one semester, I would surely watch this.

ChatGPT is not an A.I. in itself, it's an A.I for chosing correct answers from the wiki, StackExchange, blog posts and others, its A.I. is to determine which one of them could be the correct answer, that's it. It would be the same as if you google these questions and got different high voted answers, but the difference here ChatGPT is more likely to pick the correct one for you.

@gasdive

Жыл бұрын

No, that's not how it works at all. It's not copying existing works. It's not just a filing system.

@hqcart1

Жыл бұрын

@Farb S so when i ask chatgpt to write a js code that it has no database to lookup from, it will write that code? dude, everything in chatgpt is online, they even said it was fed until 2021, all what its A.I does is ranking the answers.

Despite the economic downturn,I'm so happy☺️. I have been earning $ 60,000 returns from my $7,000 investment every 13days.

@EvieRandall

Жыл бұрын

This must be an investment with Mrs Georgette Wong

@Kathryn.J.Bond.

Жыл бұрын

Mrs Georgette is the right person to start trading cryptocurrency with.. she knows her way around the crypto world.. she has been helping me increase my investment every day for over months..

@KaiWhitehouse

Жыл бұрын

That's amazing, I have been hearing about mrs Wong. and her genius mind in the crypto market, please how did you earn such good amount?..

@DylanLambert531

Жыл бұрын

Actually I trade cryptocurrency on a platform, with assistance from their top crypto experts. Mrs Georgette is my professional assistant, I have been trading with her for 8 months now... I've really made a lot from her strategies in trading of cryptocurrencies.

@StellaCallister991

Жыл бұрын

I think I'm blessed because if not I wouldn't have met someone who is as spectacular as expert Mrs Georgette I think she is the best broker I ever seen

I used this for an essay but I didn’t copy paste anything. I got examples on what to write about and then searched it myself. I then asked further about the topic and again searched manually. You can also ask it for both grammatical and factual inaccuracies once you’re done which is cool. I didn’t change any of the factually inaccurate things as this wasn’t very important for my essay and I didn’t trust it that much, I was low on time as well. It made some sentences easier to understand. As a non native speaker that was very helpful. It’s like someone else reading it and pointing out things. It saved me so much time, and honestly in this case I would not have done this good of a job writing the essay in the same amount of time I had. I had other exams that week.

Very informative, thank you for your time and effort.

First

@Imsosappy

Жыл бұрын

You truly are a hero

@brentthompson6601

Жыл бұрын

😎

@ianhopcraft9894

Жыл бұрын

Damn you blackrack !😆

@brentthompson6601

Жыл бұрын

But I was first because I called it in my head lol 😂.

Someone as far ahead of the curve in preparing for this new challenge like yourself can easily focus full-time to help brace the masses for this massive disruption coming this year with gpt-4. Good luck to us all in learning adaptation at warp speed. 🤞

Thank you for having such a great mindset as a professor!

Really enjoy the content on Cool worlds, I have by my own admission a brain that is now days resembling swiss cheese. I have some questions that have always never been able to be answered correctly with regard to Space/Time.

I love you so much! Staying thoughtful and staying curious!

Cool! These language models are usually a bit better if you ask them to provide the explanation first and then the answer, rather than the other way around like ChatGPT did here.

ist just amazing how this IA does all the things and make's the answers they must have simulated this so much.

As a PhD graduate I see ChatGPT positively. It's about using it as a "shortcut" tool. If the student is good enough to have 70% at the exam, chatgpt will fill the gaps and the student will be able to tell the small mathematical mistakes and think logically about the "too small unrealistic number" for example. Tools like ChatGPT will make our job/work/research much quicker. I see it as a higher abstraction thinking. Imagine low abstraction programming languages that used to take us 10 lines to calculate `a+b` and nowadays we can do this in 1 line with modern programming languages. In low abstract languages you need to write 50k lines of code for something that could be solved in 100 lines of modern code. So we can accomplish much greater results in much less time. I see ChatGPT as a higher abstraction thinking - you want to accomplish certain goal in 1 month, but with its help you accomplish it in 5 days - then you have 25 days to accomplish something 6 times bigger. This would help the technological progress, scientific research, everything! Obviously it relies on information that we already have, but this will save us a ton of time in researching the topics we need to research!

Great editing quality as always

I feel that your interpretation of chat GPT is accurate I also feel you know how to appropriately ask questions better now.

I can't get enough videos, content, opinions, and thoughts about, ChatGPT! I've been playing with all the public AI chat bots, image gens, up-scalers, TTS, etc. for a few years now and I just find it fascinating. I've already found practical and fun uses for most of it, but I can't help feel that little bit of unease. There's no stopping it, and I think ChatGPT is sort of the inflection point for a lot of people *realizing* that this tech is something that'll affect everyone (even though ML has already been doing that for quite a long time).

Thank you for another great video brother

This was a fantastic video that brilliantly highlighted the problem we are facing as educators. First, I completely agree with you that the old form of closed book memory testing exams is pointless now. I’ve long argued that creativity is what we should be encouraging. Let students have the information on front of them. It’s what they do with it that matters in the real world. And that perhaps where social sciences and humanities perhaps have an advantage over the natural sciences in tackling the challenge of ChatGPT. It’s possible to set assessments that would really challenge something like ChatGPT, not because the questions are too hard, but because ChatGPT is too good at providing standard pro forms answers! For example, I teach a course on Britain and Europe. For the final assessment, I asked my students to produce a 15 minute video on the following: ‘When You return home, how will you explain Brexit to your friends and family?’ I want to see how they interpret the question, how they structure their approach in a very human and personal way, and then how they present it. In other words, what we need to do is assess students on their interpretative and communication skills, linked to the hard factual knowledge. We have to get away from just having written answers. It’s certainly a challenge, especially for the natural sciences. But it’s a whole new world. We educators need to keep up with the working environment our students will face and prepare them for it. And that includes dealing with the inevitable impact of a work environment where AI will provide a lot of the answers!

The 9000 series is the most reliable computer ever made. No 9000 computer has ever made a mistake or distorted information. We are all, by any practical definition of the words, foolproof and incapable of error.

Wow what a fascinating idea for a video. There sure has been alot of buzz lately around things having to do with AI, and especially GPT-3. It's really interesting that it seems to struggle with math-specific questions. I would tend to think that questions that have definitive and certain answers like Math would be something that an AI would excel at.. This is just really fascinating to me. Not an academic person or student here. Just a guy who is retired and finds the Cool Worlds videos really super interesting.

I think we're going to have to move from a model of essays and getting the right answer and into a world where teachers/TAs have to sit down and have a conversation with students as they work through the problems to make sure they know how to think. It would be similar to an oral exam in language courses but open book/open notes. Grading rubrics would be based on using the correct laws/relationships, applying them properly, handling units, doing the calculation correctly, etc. Right answers are something that will eventually be easily referenced on any computer, it's the thinking process behind the answers that you're training and you'll need to test for. It's going to be a lot more difficult and time consuming for teachers/TAs, but it's the only way I can see forward outside of preventing students from using computers on any assignment or test.

@maalikserebryakov

Жыл бұрын

50IQ solution

This is the first time I've really heard details about chatGPT. What's really surprising and extremely impressive to me is how it is able to understand the context of the questions, do its 'research', and apply what it found to answer your questions. It even goes on to explain the concepts and how it arrived to the answer. That blows my mind. For instance, in the last question @ 23:06, the question starts off with "If an exoplanet transit has a depth of 0.003...", ChatGPT is smart enough to understand terms like "depth" that have a specific meaning in the context of the question/subject matter. It doesn't know it's taking a test for astronomy course! How is it able to determine the context so well that it can look up information to attempt to answer the question? I know it gets the question wrong, but it clearly understood the nuances. It just absolutely blows my mind that it can look up information to attempt to answer complex questions. it must be able to determine what is accurate and inaccurate information during its research process. This takes humans how long to do to just build the fundamental knowledge in subjects to get to these higher level questions in university, yet this ChatGPT can do it seconds (I assume basically instantaneously). Serious mind-fuck

Having just used Chatgpt I, as a professor in astronautics, am now going to require more specific detail from students' exams and homework submissions knowing that would be a potential downfall area in using AI as you so succinctly demonstrated in your video. Thanks!!