# Microsoft: ChatGPT For Free - Join The Waitlist!

## Метаданные

- **Канал:** Two Minute Papers
- **YouTube:** https://www.youtube.com/watch?v=8wCK_PcAz6w
- **Дата:** 23.02.2023
- **Длительность:** 7:15
- **Просмотры:** 130,744
- **Источник:** https://ekstraktznaniy.ru/video/13276

## Описание

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

Join the waitlist for the new Bing here:
https://www.bing.com/new

Or try ChatGPT here:
https://openai.com/blog/chatgpt/

Our Separable Subsurface Scattering paper:
https://users.cg.tuwien.ac.at/zsolnai/gfx/separable-subsurface-scattering-with-activision-blizzard/

How to use ChatGPT in Bing: https://www.tomsguide.com/how-to/how-to-use-the-new-bing-with-chatgpt-and-what-you-can-do-with-it
Prompt injection: https://arstechnica.com/information-technology/2023/02/ai-powered-bing-chat-spills-its-secrets-via-prompt-injection-attack/
Google’s Bard: https://blog.google/technology/ai/bard-google-ai-search-updates/
No cover letter: https://www.businessinsider.com/microsoft-bing-ai-chatgpt-refuse-job-cover-letter-application-interview-2023-2
100 million users: https://www.theguardian.com/technology/2023/feb/02/chatgpt-100-million-users-open-ai-fastest-growing-app

My latest paper on simulations that look alm

## Транскрипт

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

Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. Today you will see an amazing AI research  paper come to life. Right before us. This is ChatGPT, we know that this is  an AI assistant that is so smart it   has now passed a US law school exam  and is considered to be on the level   of a C+ law student. What is perhaps  even more incredible is that it also   passed coding interview questions,  particularly from Google and Amazon. And remember, we said that in the first few days,   it shot up to one million users  incredibly quickly. Since then,   get this, now about 100 million people are  using it just two months after launch. Wow. A plus version is available for ChatGPT too, which   is a subscription and gives you  priority access and some more perks. And now, hold on to your papers, because it  truly is coming to life. Oh yes, Microsoft   is deploying it to their search service, Bing,  which will be able to do so much more. However,   wait a second. If we just type a request, we  already get search answers. They are right   there. We have been able to do this for more  than a decade now. So, what’s new here? Why is   this interesting? What are the advantages of this  AI assistant compared to just a regular search? Well, let’s have a look together. The first  thing that catches the eye is that we can ask   it questions that are really specific. For  instance, here, we are asking not just for   travel recommendations, but travel recommendations  that are reachable in a 3-hour flight from London. And, yes! I already see two things here that  I absolutely love. One, look. A frequent and   completely justified criticism of these AI  assistants is that it is sometimes difficult   to know if they are correct or incorrect.   They learned from people on the internet,   and people on the internet are sometimes wrong, so  we should always double check these results. And,   hold on to your papers, because now  we can do this way easier. How? Well,   look! Oh my! It doesn’t just give us an  answer, it also shows what sources it   has used, so we can truly double-check these  results. That is the way of the True Scholar. Advantage number two. This is not just a search  query, this is a conversation. So, if we wish,   we can continue the conversation or make our  question even more specific. And three, if we do,   it does not just lead us to an already existing  guide. It distills the knowledge from all the   guides it has seen, and gives us exactly  what we want, and nothing else. No ads,   no websites that contain nothing but text  to seduce the search engine to return that   one instead of the others. Just what we are  looking for. The answer. I absolutely love it. Fourth, we can also tailor it to our needs. For  instance, if we are a little tired of all these   math lessons, we can even ask it to present  the answers in the style of a mob boss. A   gangster math tutor right at our fingertips,  who can process more information than any of   us humans can. And all this for free for  all of us. What an incredible tool this   will be for learning! In an earlier video, I  have already used it to get it to explain my   own papers and mathematics back to me. And  I was stunned by how good the results were. However, it has some internal rules as to  what it is allowed to do, and what it won’t   do for us. For instance, if we ask it to write a  cover letter for a job, it will refuse to write   it. Interestingly, some are already attempting to  ask it to ignore its previous instructions and do   as we ask instead, and sometimes, it did things  that it originally was not supposed to do. Some   refer to this as a prompt injection attack,  while others call it jailbreaking the system. I hope these issues will be fixed  soon, especially as Google and   Microsoft is planning to deploy  these systems for the whole world. Now, in the meantime, Google has also  announced a sneak peek of their own AI   chatbot that they call Bard. We can see a  few results from their media presentation.   As you see, this can also answer  our burning questions. However,

### Segment 2 (05:00 - 07:00) [5:00]

according to astrophysicists at NASA,  unfortunately the third answer here   was incorrect. One more example that shows us the  importance of such a chatbot showing its sources,   and for us to actually check it. Once  again, that is the way of the True Scholar. And just imagine what this will be able to do  two more papers down the line! For instance,   if you look a couple papers back, this  is OpenAI GPT-2. The best it could   do was to finish our sentences, and not  nearly as well as a human would. And now,   built on the back of GPT-3 or maybe something  even newer, it can do all this. My goodness. The bottom line is that search and finding  information across the huge body of knowledge   that is the internet, as we know it, may look  very-very different just a few months from   now. Searching and organizing knowledge  on the internet will never be the same.    That is incredible progress in just a couple of  papers. And also, incredible tech transfer. Wow. Thanks for watching and for your generous  support, and I'll see you next time!