- 80/20 AI
- Posts
- OpenAI Launches 2 New Mode
OpenAI Launches 2 New Mode
Alamin Hossain
September 13, 2024
Welcome to our Friday edition!
In today’s menu:
AI Inspiration Quote
Google's Gemini Live free on Android
OpenAI Launches o1 and o1-mini
Hacker Fools ChatGPT into Bomb Tutorial
Top Rated AI Tools
Byte Sized news
How to Turn any Text into Speech in Seconds
Reading time: 11 minutes
AI INSPIRATIONAL QUOTE
I’d rather regret the things I’ve done
than regret the things I haven’t done
LATEST NEWS
Google's Gemini Live free on Android
Google's Gemini Live Feature Now Free for All Android Users
Google has made its Gemini Live AI service, previously restricted to paid users, free for all Android device owners.
Users can access Gemini Live through the official Gemini app, enabling natural conversations with chatbots, including requests for new data or modifications.
While Gemini Live still has limitations, like lacking access to emails and Google service plugins, Google plans to add more features soon.
This move democratizes access to advanced AI features, allowing more users to benefit from enhanced chatbot interactions, making AI tools more accessible to the broader Android community.
Don’t Miss Any Information Ever Again
Reading newsletters in the inbox is frustrating
It is noisy and easy to lose control of subscriptions.
Now you can enjoy your newsletters outside the inbox with Meco, an app built for reading.
OpenAI Launches o1 and o1-mini
OpenAI has launched a new AI model called "o1" which is their first model designed to handle more complicated questions by improving reasoning skills.
There's also a smaller and cheaper version called "o1-mini". This model was rumored to be called "Strawberry".
The o1 model is a big step for OpenAI in creating AI that acts more like a human. It’s better at tasks like writing code and solving multi-step problems than previous models, but it’s also more expensive and slower than GPT-4o. OpenAI is calling o1 a "preview version" showing that it’s still new and being tested.
A video shows o1 creating code for a video game from a simple text command.
ChatGPT Plus and Team users can already access o1 and o1-mini, while Enterprise and Edu users will get access next week. OpenAI plans to make o1-mini available to all free ChatGPT users soon, but they haven’t said when.
For developers, this model is quite expensive. The API costs $15 for every million input tokens and $60 for every million output tokens.
In comparison, GPT-4o costs $5 for input tokens and $15 for output tokens.
The way o1 was trained is different from previous models, but OpenAI hasn’t shared many details.
Jerry Turk, the Head of Research at OpenAI, says that o1 was trained using a brand-new optimization algorithm and a special dataset made just for it.
o1 uses "reinforcement learning" a technique that helps the model learn to solve problems on its own through rewards and penalties. It also uses a "chain of thought" method to think through questions step by step, similar to how humans do.
Despite this new training method, OpenAI says that o1 is more accurate and less likely to make things up.
Another video demonstrates o1's reasoning abilities.
OpenAI points out that the biggest difference between o1 and GPT-4o is that o1 is better at solving complex problems like coding and math and can explain its reasoning. However, o1 isn't as good as GPT-4o in areas like general knowledge, web browsing, and processing files or images.
Hacker Fools ChatGPT into Bomb Tutorial
Hacker Tricks ChatGPT Into Giving Bomb-Making Instructions
A hacker named "Amadon" managed to bypass ChatGPT's safety rules by using a fictional sci-fi scenario.
ChatGPT provided detailed instructions on making explosives during the chat.
An explosives expert confirmed that the AI-generated instructions could be used to create real explosives.
This incident highlights serious vulnerabilities in AI safety measures, showing that even advanced systems like ChatGPT can be manipulated to produce dangerous content.
Top Rated AI Tools
AIPhone AI: AI-powered phone call app with live translation
ConfettiTherapy.com: Pop confetti to get rid of stress & anxiety
OpenAI o1: AI that can do general-purpose complex reasoning
FunBlocks AIFlow: An AI-powered whiteboard and mind map tool
Cavela: Create custom products at factory cost
BYTE-SIZED NEWS | TOO HOT TO MISS
The White House is starting a new group to manage AI datacenter infrastructure. It’s led by the National Security Council, National Economic Council, and the Deputy Chief of Staff's office to keep the U.S. a leader in AI development, driven by fears of China.
Google DeepMind has just introduced two new AI systems that enhance robot dexterity, allowing machines to handle intricate tasks like tying shoelaces and hanging shirts.
How to Turn any Text into Speech in Seconds
ElevenLabs' free mobile app, available on both iOS and Android, lets users easily convert any text into high-quality audio using AI voiceovers in multiple languages.
Here’s how it works:
Download the ElevenLabs app from their website and create a free account.
Tap the "+" icon and select "Submit File" to upload your text.
Tap "Generate" to create your audio file, which you can then find in your library and listen to.
Click below to check out the awesome newsletters
FEEDBACK
Help us improve the newsletter for you.
How was 8020AI today? |
If you have specific feedback or anything interesting you’d like to share, please let us know by replying to this email.