# OpenAI's ChatGPT: It Can Do What? 🤯

## Метаданные

- **Канал:** Two Minute Papers
- **YouTube:** https://www.youtube.com/watch?v=LVcxmmXNaj4
- **Дата:** 15.02.2023
- **Длительность:** 9:07
- **Просмотры:** 84,607
- **Источник:** https://ekstraktznaniy.ru/video/13284

## Описание

❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambdalabs.com/papers

MBA exam (paper!): https://mackinstitute.wharton.upenn.edu/wp-content/uploads/2023/01/Christian-Terwiesch-Chat-GTP-1.24.pdf

Check out ChatGPT here:
https://openai.com/blog/chatgpt/

My latest paper on simulations that look almost like reality is available for free here:
https://rdcu.be/cWPfD 

Or this is the orig. Nature Physics link with clickable citations:
https://www.nature.com/articles/s41567-022-01788-5

Law school exam: https://techxplore.com/news/2023-01-chatgpt-bot-law-school-exam.html
Google code interview: https://www.pcmag.com/news/chatgpt-passes-google-coding-interview-for-level-3-engineer-with-183k-salary
Amazon code interview: https://www.businessinsider.com/chatgpt-amazon-job-interview-questions-answers-correctly-2023-1
Assistant: https://twitter.com/justLV/status/1621253007492141056
Summarizing terms and conditions: https://twitter.com/uttamg911/status/1618751144226099201/photo/1


## Транскрипт

### Segment 1 (00:00 - 05:00) []

Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. Today you are going to witness perhaps history  in the making. You see, we have some amazing new   results with OpenAI’s ChatGPT, which is an AI  assistant that has now passed a US law school   exam and is considered to be on the level  of a C+ student. What is perhaps even more   incredible is that it also passed coding interview  questions, particularly from Google and Amazon. That sounds like one more step towards the  human-level intelligence for an AI that   just a few years ago, most people  thought will never be possible. Now, I have bad news and good news. The bad news  is that these are not results from a rigorously   peer reviewed paper, so we can’t pop the hood  and look inside to verify these claims. However,   good news, ChatGPT also took  an MBA exam with a professor at   Wharton business school at the University of  Pennsylvania, and we have a paper for that,   so we can look at the exact questions and  answers it has given, and an evaluation of   those answers as well. And the results are I  think, incredible. In short, this work is an   amazing leap forward, it can do absolutely  magical things, and then, as we saw earlier,   fall on its face on the simplest questions.   Now, hold on to your papers, because overall,   it seems to be a B to B- MBA student. Which  sounds incredible, but it gets even better,   because ChatGPT is no ordinary student. ChatGPT  is not just a student that can answer a few dozen   questions a day. It is a student that can answer  more than 10,000 questions nearly instantly. Don’t forget, it also knows about so much  more than just business administration,   it knows about physics, art,  programming, literature,   and can recall and build on a larger body  of human knowledge than any of us can. Actually, let’s have a look at 4 more amazing  and recent examples of what it can do. One, it can write a little computer game where  we can fly around in a little virtual world,   and we can even control the camera with our mouse.   Yes, ChatGPT is only capable of outputting text,   but there are ways to convert text information  into visual information. And one excellent   example of that is computer code. This computer  code can be compiled into a computer game. And to get all this, we just need  to ask. In goes one text prompt,   and out comes a computer game.   Clearly, a very simple one,   but just imagine how incredible of a learning  tool this will be for a new generation of   Fellow Scholars. Some of which you will see  at the end of this episode. You’ll love it. Two, it can also become an incredible personal  assistant. Look, it checks our e-mail, okay,   we expected that. Then, it tells us  how many unread emails we have. Okay,   we expected that too. And now…listen.   Whoa! It also summarizes the contents   of these unread emails. How cool is  that? A true AI personal assistant. Three, here is a brilliant application that  ties multiple things together. For instance,   if it has enough knowledge to be an okay  law-student and it is also an amazing assistant,   then have a look at this. A personal  assistant that summarizes the terms and   conditions. We have all been asked to accept  a huge wall of legal text that no one reads,   and even if a poor soul is asked to read  it, that person has a very small chance of   actually understanding it. However, that  is no more. ChatGPT comes to the rescue   and makes these readable for ordinary  humans in a hurry. I absolutely love   this one. And it can summarize Youtube videos as  well. I wonder what it would say about this one. Four, we can also create more incredible  combinations. For instance, we can ask   ChatGPT a question, and have it answer in  the style of our favorite people. Here,

### Segment 2 (05:00 - 09:00) [5:00]

this Fellow Scholar asked it to choose  from Ronaldo and Messi in the style of   Jesse Pinkman from Breaking Bad. Who is the  better footballer? I thought about reading   the answer to you, but you know what?   We can even have a different learning   algorithm clone the actor’s voice and use  it to answer himself. How cool is that? And, plus one because I can never resist. Here,  it was asked to describe how a Scholar should   hold on to their papers, and the amount of  detail in the answer did not disappoint. And,   yes, you are seeing correctly, the author  was Chesto. Yes, THAT Chesto, the Legendary   Scholar who earlier, showed you the reference  motion for holding on to your papers properly. And as promised, I would absolutely love to  show you a heartwarming message from some of   our younger Fellow Scholars. Look at this. The  message given with this image was as follows:  “My cute kids enjoying another two minute  papers episode over our weekly Saturday   lunch. Whenever you say "hold onto your  papers" they all slam a hand on the table. ”  And…yes! They are indeed showcasing  the proper technique. Also,   look at those grins! Learning something  new and improving themselves while they   are eating healthy veggies.   Loving it. Go Mini Scholars! I love my job talking about these papers, and  it is incredible to do it for such an amazing   and receptive audience like you Fellow  Scholars. I have to pinch myself every   day that I get to do this. Thank you so much!   And if you have similar ceremonies at home,   make sure to keep them coming either here or on  Twitter. What an incredible heartwarming message! And, one more thing. Whenever you see these news,  always keep the First Law of Papers in mind. The   says that research is a  process. Do not look at where we are, look at   where we will be two more papers down the line. I  see so many examples of other people reading these   news and just thinking about what is now, not  what will be soon. So, what will be soon? Well,   two more papers down the line, I am sure that  it will know our preferences quite well and   will summarize emails and papers exactly the  way we like it. And it will be able to do so   much more. So, what do you think? What would you  use this for? Let me know in the comments below! Thanks for watching and for your generous  support, and I'll see you next time!