OpenAI's ChatGPT: It Can Do What? 🤯

Dear fellow scholars, this is Two Minute Papers with Dr. Zsolnai. 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 a full 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. And 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 four more amazing and recent examples of what it can do. 1. 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. 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. 2. It can also become an incredible personal assistant. Look, it checks our email. Ok, we expected that then it tells us how many unread emails we have. Okay, we expected that too. And now listen.

Amy Lee says the project deadline was pushed back to next week and that she likes the new direction. Daniel Abrams has sent you a Dropbox link to upload your documents for 2022.

Whoa. It also summarizes the contents of these unread emails. How cool is that? A true AI personal assistant.

3.

Here is a brilliant application that ties multiple things together. For instance, if it has enough knowledge to be an OK law student and it is also an amazing assistant, then have a look at this. A personal assistant that summarizes 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. 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. 4. 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. 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 this answer to you, but you know what? We can even have a different learning algorithm clone this actor's voice and use it to answer himself.

Yo, yo, yo. Look man, it's a tough call, you know what I'm saying? They're both insanely talented and they've both accomplished so much in their careers. But you know, if I had to choose, I'd probably lean towards Messi.

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 Cesto. Yes, that Cesto, the legendary scholar who earlier showed you the reference motion for holding onto your papers properly. And as promised, I would absolutely love to show you the a heartwarming message from some of our younger fellow scholars. Look at this. The message given with this image was as my cute kids enjoying another two minute papers episode over our weekly Saturday lunch. Whenever you say hold on to 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 first law of papers 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 this 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 so much more. So what do you think? What would you use this for? Let me know in the comments below.

If you're looking for inexpensive cloud GPUs for AI, Lambda now offers the best prices in the world for GPU Cloud compute no commitments or negotiation required. Just sign up and launch an instance and hold on to your papers. Because with Lambda GPU Cloud you can get on demand a 100 instances for $1.10cent per hour versus $4.10 per hour with AWS. That's 73% savings. Did I mention they also offer persistent storage? So join researchers at organizations like Apple, MIT and Caltech in using Lambda Cloud instances, workstations or servers. Make sure to go to lambdalabs.compapers to sign up for one of their amazing GPU instances today.

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