- Everyday AI
- Posts
- Thanks A Million Celebration - How well do you know ChatGPT & other LLMs?
Thanks A Million Celebration - How well do you know ChatGPT & other LLMs?
First look at Meta's Llama 3.1, win ChatGPT for a year, OpenAI gets questioned by U.S. Senators and more!
👉 Subscribe Here | 🗣 Hire Us To Speak | 🤝 Partner with Us | 🤖 Grow with GenAI
Outsmart The Future
Sup y’all 👋
That was fun AF! (And I’m a little sleepy.)
Today, we decided to kick off our ‘Thanks A Million’ campaign in style, by offering up $1,000 to anyone who could ace our Large Language Model test live and in real-time. (Keep reading to see the questions, answers and results from today’s live show.)
Why did we do this?
Thought it would be fun. (It was.)
Hopefully keep y’all sharp and AI smart. (Mission accomplished.)
Celebrate our upcoming milestone of hitting a million podcast downloads, hence… the ‘Thanks A Million’ campaign launch.
We’ll have A LOT more on this in coming episodes and newsletters.
Just know — you should go sign up now and enter to win a Free Year of ChatGPT (or your fave LLM subscription.)
And we’ll be adding new prizes to the prize pool in the coming days/weeks.
Keep an eye on the Thanks A Million section here and on our site for more.
Seriously appreciate y’all.
✌️
Today in Everyday AI
9 minute read
🎙 Daily Podcast Episode: We’re celebrating our upcoming 1 million podcast download milestone with a livestream ChatGPT and LLM quiz and a $1,000 cash prize to anyone who aced it in real-time! Think you have what it takes? Give it a listen. (No prizes for after the fact, but it’s fun to quiz yourself!)
🕵️♂️ Fresh Finds: A letter from Meta’s CEO, Meta AI’s new image feature and why Target employees hate their new AI chatbot. Read on for Fresh Finds.
🗞 Byte Sized Daily AI News: Meta unveils Llama 3.1 and its 405B model, OpenAI adds FREE! finetuning for its GPT-4o Mini model, Adobe adds new AI features and U.S. Senators have questions for OpenAI. For that and more, read on for Byte Sized News.
🚀 AI In 5: Meta’s biggest AI model to date is here! Do we have a new LLM king? Or is it just hype? We take a look. See it here
🧠 Learn & Leveraging AI: Here’s an overview of what went down during our livestream quiz! Keep reading for that!
↩️ Don’t miss out: Did you miss our last newsletter? We talked about Meta launching its biggest Llama 3 model, NVIDIA unveiling a new Chinese chip and OpenAI avoiding another lawsuit. Check it here!
Thanks A Million Celebration - How well do you know ChatGPT & other LLMs? 🤔
We’re about to hit 1 Million podcast downloads, a milestone that’s pretty rare.
Wooo! 🥳
To celebrate, we gave our livestream audience a chance to bring home a bag of cash. We ran a LIVE audience quiz on today's livestream that covered ChatGPT and other LLMs.
Think you got what it takes?
Follow along and put your knowledge to the test! (Too late to win the loot, but you can quiz yourself as you listen along!)
Join the conversation and ask Jordan questions on AI here.
Also on the pod today:
• ChatGPT and Other LLMs Quiz 🧠
• Discussion on AI Models 🗣
• Insights on AI Models 🧐
It’ll be worth your 1 hour:
Listen on our site:
Subscribe and listen on your favorite podcast platform
Listen on:
Here’s our favorite AI finds from across the web:
New AI Tool Spotlight – Invue Prep provides mock interviews with AI, Arcane is an AI content marketing platform and HackerPulse gives you the ultimate dev profile.
Trending in AI – With Meta’s release of Llama 3.1, CEO Mark Zuckerberg has released a letter on open source AI.
Big Tech - Meta AI received a new feature called ‘Imagine me’ which takes your selfies and creates reimagined images based on your prompt.
Future of Work – Recently Target rolled out a new AI chatbot to help employees and apparently, they hate it.
AI in Education – A new study shows that a majority of recent grads wish they were taught AI in college.
1. Meta Unleashes Llama 3.1: The New AI Heavyweight 🤯
Meta just dropped its latest bombshell with Llama 3.1, claiming it’s the biggest open-source AI model ever, boasting a whopping 405 billion parameters. CEO Mark Zuckerberg believes this game-changer will surpass ChatGPT as the go-to AI assistant by year-end, with features like a new image generator that captures your likeness—talk about a selfie upgrade!
Meta’s also teaming up with mega-companies like Microsoft and Google to get this model into the hands of developers everywhere, proving that open-source is the future of AI.
2. Adobe Unleashes New AI Features in Illustrator and Photoshop 🧑🎨️
Adobe is enhancing its design software with exciting new generative AI features in Illustrator and Photoshop, streamlining creative workflows. The Illustrator update introduces the Firefly Vector AI model, enabling users to generate intricate vector shapes through text prompts, while the new Dimension tool adds essential sizing information to projects.
In Photoshop, the Generate Image feature and the newly available Selection Brush tool allow for improved image detail and effortless object selection.
3. U.S. Senators Put OpenAI on Blast Over AI Safety Concerns 🤔
Five prominent U.S. Senators just dropped a letter to OpenAI’s CEO, Sam Altman, grilling him on the company's commitment to AI safety amid rising concerns about cybersecurity and transparency. They want answers on whether OpenAI will allocate 20% of computing resources to safety research and how they handle employee whistleblower protections.
This inquiry comes as AI regulation debates heat up, raising eyebrows about tech giants' accountability in shaping our future.
4. FTC Cracks Down on AI-Powered Price Hikes ❌
The Federal Trade Commission is throwing shade at eight companies—including Mastercard and JPMorgan Chase—for using AI to spy on consumers and charge them different prices based on their personal data.
FTC Chair Lina M Khan is sounding the alarm, saying, "Firms harvest Americans’ personal data and put people’s privacy at risk," as the agency digs into the shady practice of surveillance pricing. The investigation could reveal how these companies are manipulating prices to squeeze more cash from unsuspecting customers.
5. Meta AI Gets Updates and Goes Global 🌏
Meta is expanding its AI assistant to 22 countries, including Argentina, Mexico, and Colombia, with new languages like French, German, and Hindi now available. Now you can have Meta AI in your pocket across WhatsApp, Instagram, Messenger, and Facebook, using innovative features like the “Imagine me” prompts that turn your ideas into images.
Plus, the upgraded Llama 40.5B model is ready to tackle your toughest math and coding conundrums with step-by-step explanations and debugging support.
6. NVIDIA’s AI Foundry: Enterprise “Supermodels” 👀
NVIDIA has just unveiled its AI Foundry, enabling enterprises to craft custom “supermodels” using the newly launched Llama 3.1 collection, featuring models up to a whopping 405 billion parameters! Major players like Accenture and Uber are already on board, leveraging NVIDIA's platform to create generative AI applications tailored to their specific industries and cultures.
With powerful tools like NVIDIA NeMo and NIM microservices at their fingertips, businesses can now supercharge their AI capabilities and enhance response accuracy like never before!
7. OpenAI bombshell: Free Finetuning for GPT-4o Mini
Was Meta’s Llama 3.1 Thunder a bit too loud?
Maybe. Maybe not.
Regardless, OpenAI dropped this bombshell on us all – free fine-tuning of its GPT-4o Mini model through September 23. Includes up to 2M training tokens each day.
Customize GPT-4o mini for your application with fine-tuning. Available today to tier 4 and 5 users, we plan to gradually expand access to all tiers. First 2M training tokens a day are free, through Sept 23.
platform.openai.com/docs/guides/fi…
— OpenAI Developers (@OpenAIDevs)
7:50 PM • Jul 23, 2024
Llama 3.1 Review – A first look at Llama 405B
Meta’s biggest AI model to date is here!
Llama 3.1 is Meta’s newest open source AI model with over 405 billion parameters.
Is Meta's Llama 3.1 the ultimate AI game-changer?
We're bringing you a first-hand look at what Llama 405B can do.
Check out today's AI in 5.
🦾How You Can Leverage:
We switched it up today with a LIVE $1,000 giveaway.
While no one ACED our quiz in real-time and the $1,000 bounty went unclaimed, we had more than 600 comments and dozens of participants.
Shout out to some of our top finishers:
If you don’t wanna watch/listen and give our quiz a try for yourself, here’s all the spoilers. 👇
Question 1: What does the 'o' in GPT-4o stand for?
Answer: 1C. Omni.
Our take: Cmon. We had to start out suuuuper easy ya know?
Question 2: When did OpenAI incorporate?
Answer: 2A. 2015.
Our take: Crazy the OpenAI crew has been around almost a decade, right? Not so fresh face apparently?
Question 3: How can free ChatGPT users access the new GPT-4o mini model?
Answer: 3C. In the model switcher at the end of a prompt response.
Our take: Sneaky little feature hiding at the end there! Different layout for free and paid users.
Question 4: What company released Claude 3.5 Opus?
Answer: 4D. None of the above.
Our take: Gotcha! This was our first trick question. Always gotta keep you on your toes! Claude 3.5 Opus has not been released. (Yet.)
Question 5: What is the reported MMLU score for GPT-4o Mini?
Answer: 5C. 82%
Our take: Not too shabby for a "mini" model, eh?
Go learn all about it. 👇
Question 6: According to reports, how much more energy does a single prompt to ChatGPT use compared to a Google search?
Answer: 6B. About 10 times more.
Our take: Yikes! Might want to think twice before asking ChatGPT what the weather is...
It’s not all bad though. Big companies are using AI to fight back against the energy-sucking demands of AI. Check out how NVIDIA uses AI for renewable energy.
Question 7: What does the GPT abbreviation stand for?
Answer: 7C. Generative Pre-trained Transformer.
Our take: We promise we're not trying to put you to sleep with these technical terms! 😴
Question 8: What can free users not do in ChatGPT?
Answer: 8B. Build GPTs.
Our take: Free users can still do a ton, but building your own AI assistant? That'll cost ya!
Wanna learn how to build your own GPT? Go watch/listen to this episode.
Question 9: How does the functionality of ChatGPT differ between free and paid users?
Answer: 9E. B and C only (Only paid users can build custom GPTs and use DALL-E).
Our take: DALL-E or not DALL-E, that is the question... for paid users only! lolz
We broke down all the differences between free and paid ChatGPT here.
Question 10: What is the approximate size in parameters of the GPT-4 model?
Answer: 10D. 1.8 trillion parameters.
Our take: That's a LOT of parameters inside that big ol LLM head. No wonder it’s smarter than us!
Question 11: What is the actual context window of GPT-4o inside of ChatGPT Plus?
Answer: 11B. 32,000 tokens. (128K is inside the API only.)
Our take: Don't let the marketing fool you - we always test these things ourselves!
If you’re reading this wondering ‘what the heck are tokens?” We have the answer.
Question 12: Inside of ChatGPT Plus, describe how GPTs work.
Answer: 12C. You can use an unlimited number of GPTs in a single chat, but only one at a time.
Our take: It's like a party where everyone has to take turns talking. How polite!
Question 13: In Perplexity Pro, which of the following models are not available for use?
Answer: 13E. GPT-4o Mini and Sonar Media.
Our take: One of these doesn't exist. Can you guess which one? (Hint: We made it up!)
We have a dedicated episode on Perplexity that breaks down the 5 things you need to know.
*Update. In late, breaking news, Perplexity JUST added Llama 3.1 405B to its lineup.
Question 14: With Perplexity Pro, what is the newest model you can use?
Answer: 14C. Claude 3.5 Sonnet.
Our take: Perplexity's keeping up with the Joneses... or should we say, the Claudes?
*Update again. In late, breaking news, Perplexity JUST added Llama 3.1 405B to its lineup. You guys see how hard it is to stay current here? lolz.
Question 15: How much does the pro version of Perplexity cost per month?
Answer: 15C. $20.
Our take: For the price of a few fancy coffees, you too can have AI at your fingertips! At this one joint, that’s like 1.2 coffees. Yikes.
Question 16: Which of the following features does Perplexity Pro offer?
Answer: 16G. B and D only (Collections and Pages).
Our take: No gems or power-ups here - this isn't a video game, y’all. Go give this video a quick watch.
Question 17: What coding languages does ChatGPT write in?
Answer: 17E. All of the above (HTML, CSS, Python, JavaScript).
Our take: ChatGPT is multilingual... in coding languages, that is!
Speaking of coding… we kinda won’t need coding for the future of web design.
Question 18: Can you upload videos into ChatGPT Plus?
Answer: 18C. Yes, ChatGPT can accept certain video formats.
Our take: Surprise! Bet you didn't know ChatGPT could handle your home movies!
Proof shorties:
Question 19: What information can ChatGPT Plus understand when you upload a video?
Answer: 19C. It can understand duration, frame rate, format, codec resolution, etc.
Our take: It's like having a tech-savvy friend who can tell you all about your videos!
Mas proof:
Question 20: Can you create a PowerPoint file inside of ChatGPT with one click to download?
Answer: 20D. A and C (Yes, in the default mode and through the use of 3rd party GPTs).
Our take: Who needs Microsoft Office when you've got ChatGPT, right?
Question 21: What is the minimum number of seats for ChatGPT Teams and Enterprise plans?
Answer: 21E. None of the above (It's actually 2 for Teams and 150 for Enterprise).
Our take: Another tricky one! We like to keep you guessing.
Question 22: By default, which large language model has no way to connect to the Internet?
Answer: 22C. Claude.
Our take: Claude's living under a rock. Can we change this already Anthropic?
Oh yeah…If you didn’t know, not all LLMs are automatically connected to the Internet. We showed you that here.
Question 23: Which company currently has the largest context window in their large language model?
Answer: 23A. Gemini.
Our take: Gemini's got a memory like an elephant - 2 million tokens worth!
We reviewed Google Gemini back in February when it was first released. Here’s what we thought.
Question 24: What are the different focus modes in Perplexity?
Answer: 24B. All, academic, writing, math, video, and socials.
Our take: Something for everyone - whether you're solving equations or scrolling through TikTok!
Question 25: Which mode inside of ChatGPT Plus does not allow you to upload files?
Answer: 25C. GPT-4o Mini.
Our take: The new kid on the block is a bit camera shy, it seems!
HOT DANG.
Your head hurt?
Or are you bursting with excitement about your newfound knowledge?
Should we do more interactive live events like this? And if you took our quiz on the pod, hit us with a reply and lettuce know how you did!
⌚
Numbers to watch
70%
A new survey shows 70% of graduates think generative AI should be incorporated into courses.
Now This …
Let us know your thoughts!
Vote to see live results
If our newsletter was sent at a more consistent time, would you read it more?Be honest. Our feelings won't be hurt |
Reply