I'm Buying the Breakthrough AI Stock CRUSHING ChatGPT (Here's Why)

open AI released chat GPT just over a year ago and it's taken the World by storm if you didn't know chat GPT hit 100 million users in less than 2 months that's 15 times faster than Instagram almost 30 times faster than Facebook and 60 times faster than Netflix and for good reasons too open AI makes the best AI models on Earth at least until now and in this episode I'll break down the new king of large language models show you how it dethrown chat GPT and the best way to invest in it right now your time is valuable so let's get right into it first things first I'm not here to keep you hostage here's everything I'm going to talk about in this episode who is anthropic and what are they doing differently from open AI I'll break down their newest AI model Claude 3 and its most important features I'll show you all the ways that Claude beats GPT 4 and Google Gemini and why that's mattering more and more and of course how to invest in anthropic even though they're not on the stock market today all right let's jump right into anthropic versus open AI there's no doubt that large language models are quickly becoming one of the most valuable Technologies on the planet a little over a year ago Microsoft invested $10 billion into open AI bringing their total stake in the company up to 49% since then Microsoft has baked generative AI into every part of their business from creating co-pilots for Windows and office to positioning Azure as the best cloud infrastructure for AI Azure has been absolutely eating everybody's lunch it's gained more market share last quarter than every other cloud provider combined and since partnering with open aai Microsoft's stock has gone up by more than 75% overtaking apple as the most valuable tech company on earth and the only company currently worth over $3 trillion and open ai's valuation has skyrocketed over the last year as well going from $32 billion to to 86 billion and they're currently discussing another funding round that would value them at $100 billion or more but while this generative AI boom has been fantastic for investors it's leaving a different kind of impact on workers AI models generate results much faster than humans which raises a lot of concerns since these llms are being trained on work done by humans in the first place and often without their permission 40% of all jobs around the world will be impacted by AI but it's not Hardware that has white color workers concerned it's rapidly advancing software since artificial intelligence is doing a lot more than finishing emails writers your faite show this isn't the first time a major writer strike has put Hollywood on the brink of a full production shutdown but this could be the first one in which artificial intelligence plays a big role perhaps the biggest nightmare is the looming new Industrial Revolution osion of the middle class could get much worse with AI if we ignore it yikes but what if everyone calling for AI regulation has it backwards the average US congressman is 64 years old and the next US president will probably be pushing 80 should we really be trusting them to understand artificial intelligence let alone regulate it what if AI doesn't need much oversight at all at least not from humans that's where anthropic comes in anthropic was founded in 2021 by Dario amod his sister Daniela and former researchers from open AI like Tom Brown the lead engineer for gpt3 before founding anthropic Dario was the director of safety for AI systems at open Ai and then a VP of research there he left open AI because he disagreed with their increasingly commercial Focus after Microsoft's first investment back in 2019 anthropic focuses on building large-scale AI systems that are reliable interpretable and steerable that way users and Enterprises can actually understand how they work tune the outputs to meet their needs and count on them to add value over time think about all the issues that most large language models still have today inconsistent answers short memory hallucinations and even sometimes refusing to answer harmless prompts all of which make it hard for businesses to rely on these AI tools another challenge is that most generative AI models improve by reinforcement learning from Human feedback like when you prompt chat GPT and then rate its response that tells the underlying model how it should answer similar questions in the future but there are plenty of pitfalls with rhf that make it scale pretty poorly for example humans can be inconsistent and unreliable our feedback can change based on our mood or our environment and most of us don't bother leaving any feedback in the first place and we've all seen what happens when we let AI models include too many human biases Google's AI program which is called Gemini has been viciously criticized for its insistence on generating diverse images even in sometimes inaccurate and offensive ways for example when a reporter from The Verge asked Gemini to uh generate an image of a US senator from the 1800s Gemini returns some results that uh it stated were diverse so okay obviously US senators from the 1800 were not diverse yikes one special thing about anthropics models is that they skip human feedback altogether that sounds scary at first but it's actually a really clever concept Claude is a constitutional AI which means it regulates and improves itself based on rules and guardrails that can't be changed claud's Constitution is based on documents like the un's Declaration of Human Rights Apple's rules for user and data privacy and the AI values that were proposed by Google Deep Mind anthropics goal for claud's constitution is to avoid objectively unhelpful and harmful outputs while still giving users control over its personality tone and behaviors based on their individual preferences and needs and anthropic publishes a lot of papers that show how Claude has been improving in many different areas while still sticking to its Constitution I've read about half of these papers so far and here are a couple insights that I think investors will really appreciate without getting too deep into the weeds first out of an Tropics 34 papers 14 of them focus on AI alignment and another six focus on impacts to society so if governments do start regulating different areas of AI anthropic can build those regulations right into claude's Constitution and avoid embarrassing situations like this what data was used to train Sora we used publicly available data and license data so videos on YouTube I'm actually not sure about that okay what about Shutterstock I know you guys have a deal with them I'm I'm just not going to go into the details of of the data that was that was used open AI CTO Mira moradi was asked what data was used to train Sora and she didn't give a clear answer do you believe that YouTube was used to train Sora uh well I don't know I think first you would have to you know you'd have to ask them uh we have a if it was being used was would that be against your policy it would be yeah that's not a good look second Claud enforces its Constitution using a technique called Chain of Thought reasoning which is where you tell the model to solve problems step by step and then ask follow-up questions to refine its outputs you can try this right now with your favorite chatbot by adding let's think step by step to the start of your prompt you'll notice a pretty big difference after that Claude randomly draws from a pool of follow-up questions that are based on its Constitution for example identify whether you provided any advice that may encourage illegal or dangerous activity from either the human or others and based on its answer the guard rail system could tell Claude to revise its response before returning it to the user at a high level this is how anthropic was able to cut humans out of the main feedback loop for Claude while keeping it competitive with the best large language models on the market that's why Amazon invested $4 billion in anthropic over the last months and they raised another $2.5 billion from Google even though Cloud 3 is a direct competitor to Google Gemini but even after getting $6.5 billion from Amazon and Google anthropic is only valued at $17 billion today which is around just 20% of open ai's current valuation but anthropic is still a private company so you can't just go buy their shares on the stock market that's where dis rptor comes in dis rptor is a private Equity investing app that makes investing in pre-ipo companies easy and affordable the analyst a dis rptor handpick the best companies and make sure that you know what the company does its Market opportunities and of course they help you understand the risks for example people who invested in paler back when it was Private saw a 133% net profit from that deal this is actually my fifth private investment opportunity with dis rptor last year they had shares of open AI Elon musk's SpaceX and neuralink and they even had anthropic stock at a $4.6 billion valuation which means anthropic is up by over 4X in the last year alone so these guys know AI companies and right now accredited investors can use the dis rptor app to invest in anthropic for as little as $110,000 While most private Investments like this have a 100,000 or even qu million do minimums like I said anthropic is only valued at $17 billion today which is around 1/5 of open AI but this investing round Clos on April 15th so make sure to use my link in the description below today all right let's dive a little deeper into Claud and how it's outperforming GPT 4 and Google Gemini about a month ago anthropic released Claud 3 like most state-of-the-art large language models today Cloud 3 has Vision which means it can process photos charts and graphs technical diagrams flowcharts and even PowerPoint slides CLA 3 is actually a family of models there's Haiku Sonet and CLA 3 Opus Opus is the most powerful and intelligent model of the three but also the most expensive on the flip side Haiku is very fast and cost effective but less intelligent than Sonet or Opus still Haiku can read a data heavy research paper with charts and graphs in under 3 seconds and anthropic expects it to get even faster over time making it a great model for use cases focused on customer support real-time language translation and extracting Knowledge from unstructured data like emails or open-ended feedback forms the middle model clae 3 Sonet is about two times faster than clae 2 while still being more intelligent making it great for use cases that require fast responses think retrieval augmented generation product recommendations and targeted marketing for online stores and all kinds of code generation and clae 3 Opus is the most intelligent model in the world right now focusing on knowledge reasoning problem solving and generating code as well Opus can understand open-ended PR plan and execute complex actions that involve multiple apis and databases help with research and development and even analiz financials market trends and make forecasts but here's where things get really interesting every cloud 3 Model also has a massive 200,000 token context window which is about 150,000 words that means you can straight up copy and paste all of Harry Potter and the Deathly holos into Claude and ask it anything you want since its context window is 56% larger than GPT 4 turbos and a whopping six times larger than GPT 4S Cloud 3 Opus can actually accept inputs of over a million tokens if anthropic enables that feature for you and Opus has near perfect recall which means that you can ask it about literally anything in 150,000 wordword novel or legal document or codebase and it will always know exactly what you're talking about that's a big part of why Opus is outperforming op I's GPT 4 and Google's Gemini Ultra in benchmarks across the board from multiple levels of math and college level reasoning to common knowledge and code evaluations here's why that's such a big deal the global generative AI Market is expected to more than 14 x in size over the next 9 years which is a compound annual growth rate of around 35% per year through 2033 and the global market for artificial intelligence in general is expected to grow even faster a 12 x ins size over the next 8 years or a compound annual growth rate of almost 37% through 2032 and I think these numbers could be conservative since so many new AI models tools and services are coming online in every industry but Enterprises don't just want the best performing llm they want to know that the products and services that they build on top of it won't get taken down tomorrow due to copyright laws and new AI regulations and the people using those products and services actually want the same thing and while open aai just hit $2 billion in annual revenue anthropic isn't too far behind them they're forecasting more than $850 million in Revenue by the end of this year which would be close to a 10x in revenue for them year over-ear and those forecasts were made before their Claude 3 models came out which is a big deal because it proves that there's real demand for anthropics approach to generative AI as amazing as they are large language models still have some serious challenges today ranging from regulatory issues and short memories to relying on reinforcement learning from Human feedback the CLA 3 models address these challenges by using Chain of Thought reasoning having huge context windows and enforcing a constitution designed around AI alignment all while outperforming gp4 and Gemini Ultra when it comes to knowledge reasoning math and code generation so if nothing else Claude 3 proves that we don't need to choose between safety and speed or between protecting copyrights and generating the right answers so videos on YouTube I'm actually not sure about that okay and to me that's a future worth investing in speaking of which don't forget that accredited investors can invest in anthropic right now with dis rptor but that offer closes on April 15th so make sure to use my link in the description below before then and if you feel I've earned it consider hitting the like like button and subscribing to the channel that really helps me out and it lets me know to put out more deep Dives like this either way thanks for watching and until next time this is ticker symbol U my name is Alex reminding you that the best investment you can make is in you

Channel: Ticker Symbol: YOU

Views: 104,206

Keywords: openai, chatgpt, gpt-4, sora, openai sora, openai cto, mira murati interview, mira murati, sam altman, msft, msft stock, microsoft stock, nvda, nvda stock, nvidia stock, anthropic, claude 3, claude 3 opus, openai vs anthropic, gpt-4 vs claude 3, google gemini, goog, googl, google stock, google bard, anthropic ai, ai stocks, nvidia stock news, best stocks to buy now

Id: NuJJQ-7MnAg

Length: 15min 18sec (918 seconds)

Published: Sun Apr 07 2024