EVERYTHING That Happened In AI In 2023

So can you believe it? 2023 is almost over and for the entire year I've done my best to keep my finger on the pulse of all the latest AI news and tools, and then turn around and make videos and share what's been going on with you. I thought since we're at the end of the year, it would be fun to make a recap of all of the most major moments in the world of AI throughout the entire year. Now, there was literally hundreds if not thousands of announcements in the AI world for the entire year, so I can't break them all down. And I there's going to be comments about hey, you forgot to mention this or you forgot that tool or this announcement, and there's going to be some stuff that I missed. But if you look at my tabs across the top of my chrome here, I've narrowed it down to roughly 45 big announcements from throughout 2023. I also don't want this video to be an hour long, so I'm going to rapid fire break down the year for you. Keep in mind that pretty much every single event that I'm going to talk about at some point in this channel's history, I made a video where I deep dived into that spec. This video is sponsored by InVideo AI, a really cool AI video tool, but I'll talk about them more in a few minutes here. So let's go ahead and talk about what happened in 2023, starting with January. In January, Microsoft invested $10 billion into OpenAI, kicking off what seems to be one of the most brilliant moves from Microsoft because since then they've been able to put all of OpenAI's tech into their products. January also saw the launch of the 11 Labs beta, giving us the first pretty good text to speech we've heard. Up until 11 labs, all of the text to speech sounded really robotic and wasn't fooling anybody. After 11 labs, people started having a hard time telling whether or not a voice was generated with AI. So much so that a whole bunch of scams and fake audio has started to come out because of technology like what Eleven Labs released in the AI art world. In January we saw instruct Pix2Pix. This was released as open source and was really the first time we were able to do image editing with AI. Up until this we were able to generate images with things like Dall E and Stable Diffusion in Mid Journey, but once this was released we were able to take the images we generated and change them, swap sunflowers with roses Add fireworks to the sky. Make his jacket out of leather. What would it look like if it were snowing? This was a huge advancement in the AI art world. Moving into February, this is when Google started to get serious about trying to keep up with OpenAI and Microsoft announcing the release of Google Bard. When Bard was released, it was exciting to see Google getting into the chatbot game along with OpenAI, but it still left a lot to be desired. Pretty much everybody that was out there experimenting with Bard was pretty unimpressed with what it could do because what we had to compare it to was ChatGPT, which at that point had become the gold standard in AI chatbots. Literally the day after Google announced Bard, Microsoft said, we're now doing Bing Chat. And not only was Bing Chat starting to roll out almost immediately because of Microsoft's partnership with OpenAI, it was using the GPT technology underneath. And we would come to find out later that it was actually already using GPT4 before we even got access to GPT4 inside of ChatGPT. In the AI art world, February was the month where we got control nets. Control. Net was a huge leap forward in AI art because it now allowed us to put our AI generated images into custom poses and positions. We were able to upload an input image and generate new images that followed the exact same pose of the original image that we uploaded. It also gave us the ability to scribble images and get more realistic output images, as well as turn existing images into depth maps and then use those depth maps for our output again. This upleveled the customizability of our AI art images. Moving into March. March was probably the biggest year for AI in all of 2023. This is when it was just announcement after announcement. During this time I was doing my best to try to keep up and I was putting out news videos almost every single day because the announcements just kept on rolling. And it started with one of the biggest advancements in the world of large language models. The release of GPT4. GPT4 was a huge leap over what were getting with the previous GPT 3.5. And the demo that Greg Brockman gave when they made the announcement blew everybody's mind, especially when he showed this off.

So I have here a nice hand drawn mock up of a joke website. Literally take a photo of this mockup and so we can actually take now this output. And here we go. Actual working JavaScript filled in the jokes for comparison. This was the original of our mockup.

And at the time we hadn't seen anything like that. Yet now I did speed up that video video to cut out some of the delay between him showing us the photo and us actually getting the output. But just what it can do was so much more advanced than what we were able to do with GPT 3.5 before that. March was also the month that we got Mid Journey version five, which was a huge leap in realism from our AI photos. We went from our AI generated images looking something like this to now all of a sudden looking like this or like this. The realism just took a huge leap with version 5. You may even remember a lot of people on the Internet getting fooled by this image that was created with Midjourney version 5. It was also March when we got Model Scope, which up until that point was the absolute best text to video we've seen. Now anybody could go and enter a prompt like an astronaut riding a horse and get a two second clip that kind of looked like an astronaut riding a horse or kind of looks like Spider man surfing. But it was obviously trained on Shutterstock content and every single video had this Shutterstock watermark on it. But it also led us to some pretty amazing memes, like this now pretty iconic meme of Will Smith just perpetually eating various spaghetti. March was also the month that Adobe entered the game with Adobe Firefly, where we could now generate AI art images for from Adobe's platform. Now did it really have a big leap and improvement over what we were getting over stable diffusion or midjourney? Not yet. But when Adobe starts rolling out Firefly into their suite of tools like Photoshop, Illustrator and Premiere and things like that, that's when Firefly really starts to get exciting. March was also the month where OpenAI announced that ChatGPT would be getting plugins extending the capabilities of what ChatGPT can do. And at launch we got plugins like Expedia, Fiscal Note, Instacart, Kayak, OpenT Shop, Klarna, Wolfram, Zapier and more. And the Demo made these ChatGPT plugins look pretty exciting. Like this demo where somebody was looking to eat vegan food in San Francisco and wanted a suggestion for.