OpenAI’s New AI Writes A Letter To Humanity! ✍️
7:07

OpenAI’s New AI Writes A Letter To Humanity! ✍️

Two Minute Papers 24.04.2022 108 380 просмотров 5 068 лайков

Machine-readable: Markdown · JSON API · Site index

Поделиться Telegram VK Бот
Транскрипт Скачать .md
Анализ с AI
Описание видео
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambdalabs.com/papers 📝 The post about GPT-3's Edit and Insert capabilities are available here: https://openai.com/blog/gpt-3-edit-insert/ ❤️ Watch these videos in early access on our Patreon page or join us here on YouTube: - https://www.patreon.com/TwoMinutePapers - https://www.youtube.com/channel/UCbfYPyITQ-7l4upoX8nvctg/join 🙏 We would like to thank our generous Patreon supporters who make Two Minute Papers possible: Aleksandr Mashrabov, Alex Balfanz, Alex Haro, Andrew Melnychuk, Angelos Evripiotis, Benji Rabhan, Bryan Learn, B Shang, Christian Ahlin, Eric Martel, Gordon Child, Ivo Galic, Jace O'Brien, Jack Lukic, Javier Bustamante, John Le, Jonas, Jonathan, Kenneth Davis, Klaus Busse, Lorin Atzberger, Lukas Biewald, Matthew Allen Fisher, Michael Albrecht, Michael Tedder, Nevin Spoljaric, Nikhil Velpanur, Owen Campbell-Moore, Owen Skarpness, Paul F, Rajarshi Nigam, Ramsey Elbasheer, Steef, Taras Bobrovytsky, Ted Johnson, Thomas Krcmar, Timothy Sum Hon Mun, Torsten Reil, Tybie Fitzhugh, Ueli Gallizzi. If you wish to appear here or pick up other perks, click here: https://www.patreon.com/TwoMinutePapers Thumbnail background image credit: https://pixabay.com/images/id-1052010/ Thumbnail background design: Felícia Zsolnai-Fehér - http://felicia.hu Károly Zsolnai-Fehér's links: Instagram: https://www.instagram.com/twominutepapers/ Twitter: https://twitter.com/twominutepapers Web: https://cg.tuwien.ac.at/~zsolnai/ #OpenAI #GPT3

Оглавление (4 сегментов)

Intro

Dear Fellow Scholars, this is Two Minute  Papers with Dr. Károly Zsolnai-Fehér. Today we are going to have an AI write a story  for us, and even do our coding exam materials.    And the coolest thing is that both  of these will be done by the same AI. How? Well, OpenAI’s GPT-3 technique is capable of  all kinds of wizardry as long as it involves text   processing. For instance, finishing our sentences,  or creating plots, spreadsheets, mathematical   formulae, and many other things. While their  Dall-E 2 AI is capable of generating incredible   quality images from our written descriptions.   Even if they are too specific. Way too specific. Now, today, we are going to play  with GPT-3, the text AI that just   gained an amazing new capability. And  that is, editing and inserting text.    And if you think that this doesn’t sound  much, well, check out these examples.

Examples

Let’s start a story. Today’s the big day! It says.   Now, the AI sees that the name of the section   is high school graduation, and it infers, quite  intelligently that this might be a message   from a school to their graduating students. So  far so good, but this was possible before too.    So what’s new here? Well, hold on to  your papers, and now, let’s do…this!    Oh yes. Change the story. Now, we don’t know why  is this the big day, but we know that the next   section will be about moving to San Francisco,  and…oh my! Look at that! It understood the whole   trick, and rewrote the story accordingly. But,  you know what? If we wish to move to Istanbul   instead - not a problem. The AI gets that one  too. Or, if we have grown tired of the city life,   we can move to a farm too, and have  the AI write out story for that. Okay, that was fantastic, but when it comes to  completion, I have an idea. Listen. How about   pretending that we know how to make hot  chocolate. Fill in the last step, which is,   you know, not a great contribution, and let the AI  do the rest of the dirty work. And can it do that?    Wow, it knows exactly that this is recipe,  and it knows how to fill that in correctly.    I love it. The put hot chocolate  part is a bit of a copout,   but you know what, I’ll take it. From the  recipe, it sounds like a good one. Cheers! Now, let’s write a poem about GPT-3. There we go!   It rhymes too, checkmark. And now, let’s ask the   AI to rephrase it in its own voice. And…yes! It is  able to do that, while carefully keeping the rhyme   intact. And when we ask it to sign it, look.   We get a letter to humanity. How cool is that? Now, onwards to coding. And this is  where things get really interesting. Here is a piece of code for computing  Fibonacci numbers. Now, translate it   into a different programming language. Wow, that  was quick. Now, rewrite it to do the same thing,   but the code has to fit one line. And here comes  my favorite: improve the runtime complexity   of the code. In other words, make it more  efficient. And there we go. Now, clearly,   you see that this is a simple assignment, yes.   But. Make no mistake, these are fantastic signs   that an AI is now capable of doing meaningful  programming work. And if you have a friend who   conducted coding interviews, show it to them. I  reckon they will say that they have met plenty   of human applicants who would do worse than this.   If you have your own stories, you know what to do.    Of course, leave them in the comments  section, I'd love to hear them. And, if you have been holding  on to your papers so far,   now, squeeze that paper because here comes the  best part. We will fuse the two previous examples   together. What does that mean? Well, we will use  the hot chocolate playbook and apply it to coding. Let’s construct the start and the end of  the code, and hand it away to the AI. And,   yes, once again, it has a good idea as to  what we are trying to do, it carefully looks   at our variable names and fills in the key  part of the code in a way that no one will   be able to tell that we haven’t been doing  our work properly. What a time to be alive!

Conclusion

So, an AI that can write a story for us, even in a  way that we only have to give it the scaffoldings   in the story, and it fills in the rest. I  love it. So cool! And it can help us with   our coding problems, and even help us save a ton  of time massaging our data into different forms.    What an incredible step forward in  democratizing these amazing AIs. So, what would you use this for? What do you  expect to happen a couple more papers down the   line? Please let me know in the comments  below. I’d love to hear your thoughts.

Outro

Thanks for watching and for your generous  support, and I'll see you next time!

Другие видео автора — Two Minute Papers

Ctrl+V

Экстракт Знаний в Telegram

Экстракты и дистилляты из лучших YouTube-каналов — сразу после публикации.

Подписаться

Дайджест Экстрактов

Лучшие методички за неделю — каждый понедельник