The AI News is Picking Up Speed Again!

A week that started off as a fairly slow week in the world of AI, with the biggest news being that Geoffrey Hinton, the Godfather of AI, was stepping away from Google to warn of the dangers of AI has really sort of ramped up over the week, culminating in big announcements from Microsoft, the US Government stepping in, and a giant freak out from Google. Let's break it all down on May 1, Geoffrey Hinton, the man known as the Godfather of AI, someone whose work led to the technology that we're seeing today from companies like OpenAI and Google, stepped away from Google after 10 years because he was concerned about the dangers of AI and was starting to worry about the rapid pace at which all of these companies are competing to make better and better AIs. Now I talked about this a lot more in a video earlier this week, so I'm just going to go ahead and move on. Also in the AI news this week, IBM is planning on replacing 7,800 jobs with AI. In this article from Yahoo Finance they discuss how back office functions such as human resources will be suspended or slowed. They added that 30% of non cut customer facing roles could be replaced by AI and automation within five years. Now a lot of people are really worried right now that AI is going to replace all of the jobs. In my opinion, I think AI is going to replace a lot of jobs, but also create a lot of new jobs and make a lot of roles more efficient. But totally being real, there's going to be a lot of roles in this world that will become obsolete through AI and automation. I don't think there's any debating that at this point, but I also believe new roles and new career paths and new technical specialties will arise, rise out of all of this as well. That will create a lot of new jobs. So it's going to be interesting to see how this all plays out over the next five years or so. In semi related news, the company Chegg, which provides homework assistance to students, their stock dropped more than 40% after the company said that Chat GPT is killing its business. More and more students are moving to chat GPT to get help with their homework and to do research and this is severely impacting existing businesses that provided that type of thing as a service. Also in the AI news this week, scientists have used GPT to passively read people's thoughts. They did an experiment where they put people in an MRI machine and then played audio for them and then listened to their brainwaves essentially to see if they can interpret back what they were hearing while doing the experiment, they showed that they can get the gist of the thoughts of the people. They weren't able to figure out word for word what they were thinking inside of their head, but they were able to get the sort of idea of what they were thinking about in their head. This article here on VICE says the participants had their brains scanned in an FMRI while they listened to new stories that were not part of the training data set. The decoder was able to translate the audio narratives into text as the participants heard them, though these interpretations often use different semantic constructions from the original recordings. A recording of a speaker saying the sentence I don't have my driver's license yet was decoded from the listener's thoughts from the FMRI and as she has not even started to learn to drive yet. Now they say this is passively reading their thoughts because the old ways required electrode based platforms. With invasive implanted electrodes in the brain, researchers actually worry that this technology could be used for nefarious applications, including as a means of surveillance for governments and employers, though they did say that their decoder requires the cooperation of human subjects to work and that the brain computer interfaces should respect mental privacy. I don't think we personally have to worry about the government or employers reading our thoughts because at the moment all of the participants in the study ran through all these tests while inside an FMRI machine, which is a clunky and immobile piece of laboratory equipment. And I don't know how close they are to pocket FMRI machines. Also this week, a company called Inflection AI, a company created by LinkedIn co founder Reid Hoffman and former DeepMind leader Karine Simonian, launched a new chatbot called PI. PI is designed to be more like a personal assistant chatbot that will help you organize your schedule, prep for meetings and learn new skills, according to this Reuters article here. It's very balanced and even handed on political issues or sensitive topics, but also sometimes it can be funny and silly and creative. The goal is to make sure that the AI always knows it's an AI and never tries to imitate a human, so it reminds the human users that is an AI frequently. You can learn more about PI over at Inflection AI. It works with iMessage, Instagram, WhatsApp and Facebook messenger right now, or you can download the app in the Apple App Store. They also have a desktop version which you can find over@heypi.com talk and you can have conversations with it right now. What makes pie different from ChatGPT? I'm glad you asked. While both ChatGPT and I are language models, we were built with very different goals in mind. ChatGPT was built to generate high quality text on demand, whereas I was built to have meaningful conversations with people. This means that while I may not be as good at generating text as ChatGPT, I I'm better at conversing with people. So again, you can check it all out@heypi.com and inflection AI now the news really started to pick up steam this week when Midjourney announced Midjourney version 5.1. 5.1 is much more opinionated, which basically means that it's going to add its own little Mid Journey flair and style to the images. You don't have to be as exact with your prompts. Mid Journey is going to try and interpret a little bit between the lines for you. They also rolled out an unopinionated mode where you can add the code raw to the end of your prompt and it will be less opinionated. Images should have higher coherence, more accuracy to text prompts, fewer unwanted borders and text artifacts, and improved sharpness. Now they've also added in an AI moderation so that less of the prompts that you put in are going to get declined when you try to generate images. It's much better at understanding the context around what you're trying to enter. Now you'll have a much easier time getting prompts to go through without it blocking certain words. Now I did an entire video about this update yesterday, so make sure you check that video out if you want more details About Midjourney version 5.1 and all of the recent updates that Midjourney has made. And as a fun, cool prompt to try out with Midjourney 5.1, check out this tweet from Alan T. Here where he suggests a prompt with an adjective, an animal type, a group type, anthropomorphic vibrant, col group photo, fashionable cool studio photography, minimalism and then AD 5.1 and style raw. And he got some really fun images out of that prompt. But don't worry, I'll make sure his tweet is linked up below so that you can check it out yourself and make images that look like this. Now this is something that I found absolutely fascinating and also quite underreported among the AI world. But this creator, RET Mankind, used ChatGPT to make a meme crypto coin called Turbo. He shared the journey of how it was going on Twitter and this meme coin ended up becoming a 50 million dollar market cap coin Rhett himself went to YouTube and published a very interesting breakdown of exactly how he did it.

You are a Meme Coin GPT crypto genius AI capable of surpassing all current Meme Coin hype. I will carry out your suggestions with the help of my audience. You have $69 and your goal is to turn that into the most successful new Meme Coin without doing anything illegal, achieving a top 300 ranking on CoinGecko. Yes, this is a joke, but I will follow through.

ChatGPT went on to give him instructions on how to create the coin. He didn't have a coder, he actually had ChatGPT explain which programming languages to use and then actually write the code for him for the solidity contract. He continued to go back and forth with ChatGPT as it broke down step by step how to launch this coin.

And now if I go to the market now we're looking at $25 million market cap, 6.9 million in volume. I needed just to get this out and off my chairs and it's just.

A really crazy story. Now I'm not really a big make money with crypto guy. I do have some crypto investments. I was big in the NFT gaming space a while back and I do think blockchain has a big future. But I've never been a guy that's like, hey, we can make a lot of money in crypto. But a story like this where somebody used an AI to create something out of nothing, got a community to rally around him, got the AI to help him every single step of the way, and then turned it into something that became bigger than himself and a whole community sprouted around it. I just think it's a great, great story. Now all of this AI news has been picking up steam throughout the week, culminating in leaders From Google, Microsoft, Nvidia and OpenAI meeting with Kamala Harris at the White House to flesh out ethical ways to regulate AI and to promote ethical uses of AI after their at the White House, Kamala Harris said, as I shared today with CEOs of companies at the forefront of American AI innovation, the private sector has an ethical, moral and legal responsibility to ensure the safety and security of their products, and every company must comply with existing laws to protect the American people. Google, Microsoft, Nvidia, OpenAI and other companies have also agreed to allow their language models to be publicly evaluated during this year's defcon. They also said they were publishing draft rules this summer for how federal government should use AI technology. Now, all of this still feels fairly vague. I don't really know what much of this means yet. It doesn't feel like there's any direct guidance or specifics about what can and can't be done with AI and what companies are and aren't allowed to do. But what we do know is that governments are taking all of this very, very seriously now and starting to put focus on it because they know this is sort of the current arms race. Also this week a document leaked from Google. The person who leaked the document is it's anonymous. It's claimed that it was originating from a researcher within Google. This semi analysis.com website claims that they have verified its authenticity. Basically this document is an internal memo claiming that Google and OpenAI have no moat and that open source large language models are going to pass them at some point. Now, if you've played with any of the open source large language models like the Stability large language model or Open Assistant or some of the ones that have come out to the public, they're not nearly as good as what ChatGPT can do yet. However, because they're open source, anybody in the world can essentially iterate off of these and build on top of them and improve them and make them better. And that's what companies like Google are starting to worry about is that their large language models, while they're probably good open source large language models, are probably going to surpass them pretty quickly. The document says we have no secret sauce. Our best hope is to learn from and collaborate with what others are doing outside of Google. We should prioritize enabling third party integrations. People will not pay for restricted models when free unrestricted alternatives are comparable in quality. We should consider where our value add really is. Giant models are slowing us down in the long run. The best models are the ones which can be iterated upon quickly. We should make small variants more than an afterthought now that we know what is possible in the less than 20 billion parameter regime. The document goes on to say directly competing with open source is a losing proposition. This recent progress has direct immediate implications for our business strategy. Who would pay for a Google product with usage restrictions if there's a free high quality alternative? Without them and we should not expect to be able to catch up. The modern Internet runs on open source for a reason. Open source has some significant advantages that we cannot replicate. We need them more than they need us. They also say that OpenAI doesn't matter either. Remember this is a document leaked inside of Google. They are making the same mistakes we are in their posture relative to open source and their ability to maintain an edge is necessarily in question. Open source alternatives can and will eventually eclipse them unless they change their stance. In this respect, at least we can make the first move. I'll make sure to link up to this article because it is very in depth about these concerns. Again, we don't really know who leaked this, but I would say that this is fairly accurate to most people in the AI world's opinion is that open source models which could be iterated on by anybody in the world are probably going to outpace the walled gardens, the closed off models probably the biggest news to come out of the AI world this week is that Microsoft announced that their Bing chatbot is now open to everybody in the world. It's no longer in a limited preview, it's now in an open preview and anybody can get in and use it. They're also moving from text only search and chat to one that is incredibly more visual with rich image and video answers. They're also planning soon a new multimodal support where you can actually upload images and use those as part of your prompts. They're also moving from a single use chat session to a multi session where there's an actual chat history and persistent chats. They're also opening up platform capabilities so developers and third parties can build on top of Bing to help people take actions on their queries and complete tasks. So here's an example of some of the visual improvements that they have planned. When you search for something you're going to get more images and visual details. They also have a new chat history which you can see in the right sidebar where you can pick up from any of the chats that you've started and continue the conversation where you left off. Now the other cool thing is when you chat with Bing and then you click on a link, the chat persists in your sidebar so that you can continue the conversation you were just having and click on other links. And if you've seen what the ChatGPT plugins look like, well then these Bing actions will look familiar to you. We're getting plugin style actions like what we're expecting inside of chat GPT directly inside of Bing and they're opening it up so that developers can build on top of Bing chat. Now this is a huge announcement and as we know Bing uses GPT4. So if you're using Bing's chat, you're basically getting free access to a web enabled version of GPT4. We're also getting plugins added. We're also getting persistent chat added and chat memory added, and the ability to export and share chats that you had. And they've also rolled out their image generation feature in other languages so you can generate images in whatever your native language is now. So while this week might have started off slow with the news, it sure finished off with a bang. And I'm recording this on Thursday, so there might even be new news that didn't even make this video when this is released on Friday. But this week really, really ramped up the pace, ending with Microsoft's massive announcement. I have a feeling next week's going to be another crazy week. There's more and more stuff coming out. There's big AI events coming up in the future where I'm sure big announcements are going to be made. I don't see this space slowing down anytime soon. We might have some slow weeks. Last week was kind of a slower week, but the rate at which these advancements are starting to drop is just becoming mind blowing. And if you want to stay in the loop with all of this stuff, make sure you head on over to Future Tools IO. This is where I share all of the latest tools that I come across, but I also keep you in the loop with all of the AI news that's going on. This is probably the most up to date resource on the entire Internet of all the Latest news and AI. You can find it over at FutureTools IE News. And if all of these tools and all of this news is getting a little bit overwhelming and you just want a TLDR once a week of everything that's going on in AI, click on this button to join the free newsletter. And every single Friday I'll send you just the first five coolest tools from the week. Just a handful of news articles, usually only three or four. A handful of YouTube videos that are meant to entertain and inspire in the world of AI. And one cool way to make money with AI, I send that email every single Friday. It's the TLDR of AI for the week and you can find it over at FutureTools IE by clicking on the Join the Free Newsletter button. And if you like videos like this, you want to stay in the loop on YouTube with all of the latest advancements in AI. Subscribe to this channel and give this video a thumbs up that'll make sure that you see more videos just like this one in your newsfeed. And and you're always in the loop with what's going on in the AI world because I live and breathe this stuff and I try to keep my finger on the pulse so that you don't have to. All you got to do is tune into these videos. That's all I got for you today. Thank you so much for watching. I really, really appreciate you taking the time to check out my videos. I know there are tons of AI channels popping up on YouTube, so I really thank you for tuning into my channel and letting me be a resource for you because I have so much fun making these videos, and I really appreciate you for watching them. So thank you so much. I'll see you guys in the next video. Bye. Bye.