AI News: We're One Step Closer To AGI This Week!

Summary

This week in AI news, OpenAI outlines five levels of AGI progress, with current tech near Level 2. A new reasoning tech, 'Strawberry,' is speculated to push towards Level 3. Controversies arise over OpenAI's employee practices, while advancements in AI-generated images, videos, and educational tools are highlighted. HubSpot offers a free AI resource bundle, and new models like GPT-40 Mini and MistrAL Nemo promise enhanced capabilities. The video also covers AI's role in forensics, e-commerce, and the upcoming Olympics.

Takeaways

• 🧠 OpenAI has outlined five levels of progress towards AGI, with current technology at Level 1 for chatbots and near Level 2 for reasoners that can solve human-level problems.
• 🔍 OpenAI is reportedly working on a new reasoning technology called 'Strawberry', which is believed to be a rebranded version of the previously mentioned 'QAR' project.
• 🤖 'Strawberry' aims to perform deep research by autonomously navigating the internet, indicating a move towards Level 3 AI that can take actions on behalf of users.
• 🗳️ Whistleblowers from OpenAI have raised concerns about the company's practices regarding employee rights and non-disclosure agreements, which OpenAI has refuted.
• 🖼️ There is speculation that OpenAI's image model, DALL-E, may have received an update, as demonstrated by clearer text in generated images.
• 🎥 Sora AI's video generation capabilities are generating excitement, though the release of other tools like Runway Gen 3 and Lum's Dream Machine have somewhat diminished the anticipation.
• 🛍️ HubSpot has released a free resource bundle for using AI at work, including flowcharts, templates, and checklists to integrate AI into various professional tasks.
• 🏫 Andrej Karpathy, former OpenAI figure, has announced a new venture, Eureka Labs, focusing on AI-assisted education.
• 📱 Anthropic's CLA, an AI chat app, has been released for Android, expanding accessibility beyond iOS.
• 🎨 Google's new app, Google Vids, is an AI-powered video creation tool integrated with Google Workspace, currently in testing with a select group of users.
• 🎵 YouTube is testing 'Music Sound Search', a feature that identifies songs based on user humming or singing, akin to the Shazam app.

Q & A

What are the five levels of AI progress outlined by Open AI?

Open AI has defined five levels of AI progress: Level one includes chatbots with conversational language capabilities. Level two involves reasoners capable of human-level problem-solving. Level three consists of agents that can take actions on our behalf. Level four is about innovators that can aid in invention by creating novel ideas. Finally, level five is about organizations and AI that can do the work of an entire organization.

What is the current status of AI in terms of these levels?

As of the script's recording, AI is at level one, with capabilities of chatbots and conversational AI. It is very close to reaching level two, which involves reasoners that can solve problems at a human level.

What is the new reasoning technology being developed by Open AI?

Open AI is working on a new reasoning technology codenamed 'Strawberry'. It is designed to not only generate answers to queries but also to plan ahead and navigate the internet autonomously to perform deep research.

What is the aim of the 'Strawberry' project?

The aim of the 'Strawberry' project is to enable AI to perform long-horizon tasks or complex tasks that require planning ahead and performing a series of actions over an extended period of time. It is intended to conduct research by browsing the web autonomously.

What controversy has arisen regarding Open AI's practices with employees?

Whistleblowers have claimed that Open AI illegally prevents employees from talking to government regulators about problems at work and removes their rights to rewards for whistleblowing. Open AI refutes these claims, stating they have a policy that protects employees' rights to make protected disclosures.

What updates have been speculated about Dolly image model?

There is speculation that the Dolly image model may have received an update, as demonstrated by a post showing an image with legible text. Previously, Dolly struggled with generating clear text in images.

What is the significance of the new demo videos from Sora?

The new demo videos from Sora showcase impressive black and white clips, indicating advancements in AI-generated video technology. These demos are increasing anticipation for the release of Sora.

What is the new AI-powered video creation app announced by Google?

Google has announced 'Google Vids', an AI-powered video creation app designed for work and integrated with the Google Workspace Suite. It is currently being tested with a select group of trusted testers.

What is the controversy about the source of training data for AI models?

There is controversy over the use of data from Uther AI, which collects data from various sources, including YouTube videos, to train AI models. Some YouTubers have noticed their transcripts being used in the training data set, raising questions about data privacy and consent.

What new model has Open AI launched recently?

Open AI has launched a new model called GPT 40 Mini. This model is designed to be more cost-efficient and faster than its predecessor, GPT 3.5, and supports text and vision in the API with future support for text, image, video, and audio inputs and outputs.

What collaboration has resulted in the creation of Mistral Nemo?

Nvidia and Mistral have teamed up to create Mistral Nemo, a 12 billion parameter model designed for on-device deployment, catering to businesses with limited internet connectivity or stringent data privacy requirements.