Google Bard: Is It Better Than ChatGPT?

Dear Fellow Scholars, this is Two Minute Papers  with Dr. Károly Zsolnai-Fehér. Dr. Papers. Today we are living the age of AI chat  assistants. That is not by accident,   these assistants are incredible, and with  browsing support, they can find deals for   us if we wish to buy a pair of headphones,  play text-based games with us that we make   up on the fly, they are excellent at  coding and can even pass the bar exam. Now I have heard some folks saying  that Google is so far behind on this,   and today you will see that this is not  the case at all. Google is coming with   guns blazing. Here is an incredible demo  of Bard, their ChatGPT-like large language   model assistant. So, what can it do? Is  it as good as ChatGPT? Let’s see together. It can, for instance, code up chess moves in  python. Although it’s python, so the task is   now as easy as it gets, it is nice to see that  it knows all the relevant libraries and concepts   to make it happen. But there is something super  interesting happening here. Look! It will tell   you what the output of this piece of code will  be. Something that ChatGPT is not that willing   to do. And it gets better - it also cites its  sources, even when it comes to writing code. Then, when done, we will be able to move  this Colab, an online notebook for running   your coding experiments. And it also has the  favorite of the Fellow Scholars, dark mode. So these assistants are mostly limited to text  information. For instance, we can ask ChatGPT   to recommend us text to image prompts, and the  results will often be fantastic. But it can not   generate an image, at least, not directly. To show  how powerful this concept is, we will now take   these ridiculously detailed scholarly prompts that  really embody the spirit of Two Minute Papers,   and to go a separate text to image AI and ask  it to generate these scholarly images. First,   Illustration of a computer program  rapidly sifting through scientific papers,   extracting key findings and condensing them into  a two-minute format. Loving it. Very imaginative.    A digital representation of a sentient AI,  depicted as a scholar, surrounded by floating   abstracts of research papers with a timer in the  background counting down two minutes. Artwork of a   futuristic AI professor delivering a rapid-fire  lecture about a complex computational topic,   with visual aids and diagrams materializing  from a holographic interface. And he is doing   it with style. So good! And these prompts  are just getting more and more ridiculous. Get this, now, a visual metaphor of a dense  scientific paper being wrung out like a wet towel,   with the drips forming simplified, easily  digestible information nuggets. Wow! And it   still delivers! And if we so desire, we can ask  for 10,000 variants of these concepts and it will   do exactly that, and very, very quickly. So just  imagine how powerful this concept of having large   language models be able to create images would  be. But the assistant itself can not output or   even show images directly. Until now! If we ask  Bard to recommend sights to see in New Orleans,   oh my. What do we have here? Look! Images are  part of the recommendations too. Note that these   images are found, and not generated, but I am  fairly confident that this is going to happen,   and quickly. But wait, it gets better,  what is that? Is it possible that? Oh yes,   that is an image input, which we can now provide  to Bard, and it will do whatever we wish,   in this case, it will write a funny  caption for it. Or as many as we wish. And as a good Fellow Scholar, if we  wish to get back and learn some more,   and perhaps go to college, and  we are interested in video games,   it can recommend colleges that offer this kind  of programme, and that is not so surprising.    The surprise comes now. And now it fires up  Google Maps and shows them to us on the map. And this is just a small example that showcases  the true potential of these assistants. And that   is, integration with other programs. And if  we wish to cram all this information into a   beautiful little table, we can do that too.   But wait, ChatGPT can do this as well, so why

is this interesting? Well, hold on to your papers  Fellow Scholars and look! We can add an additional   column to it on the fly describing whether these  colleges are private or public. And here you see   one more instance of good software integration,  where we can add it to google sheets and share   it with our friends with just a click. However,  the integration is not limited to Google’s apps,   you will be able to infuse this assistant with  the power of Instacart, Spotify, and my favorite,   Khan Academy in the next few months. Our next  video is going to be on exactly that the work   is beyond amazing, so make sure to subscribe  and hit the bell icon if you’re interested. But the list does not stop there. Remember that I  said that it would be absolutely amazing to have   the AI write prompts and generate the images in  one go? Look. Oh my! Adobe Firefly integration!    Yes! That is exactly what is going to happen  soon. What a time to be alive! And it gets better,   Bard is free to use as of now, and is available  in 180 countries as of the making of this video. And finally, there is now competition in this  space, which is really good for us, the users.    And don’t forget, all this is possible because of  the existence of these amazing research papers. In   ChatGPT’s case, GPT, and in Bard’s case, a work  called Palm. So once again, everyone is talking   about the Papers! Some might not know about it,  but they do. And just imagine what we will be   able to do just two more papers down the line. My  goodness. So, for now, let the experiments begin! Thanks for watching and for your generous  support, and I'll see you next time!