AI NEWS! | Google AFRAID of Open Source, Chatgpt Code Interpreter, AI Music & More!

TL;DR

This week's biggest AI news is the launch of DeepFloyd IF, a fully open-source image generator with built-in upscaling and inpainting that posts a zero-shot FID-30K score of 6.66, beating DALL-E 2 and Stable Diffusion on benchmarks. The roundup also covers ChatGPT's official browsing plugin and Code Interpreter, GPT-4 gaining web browsing, Amazon's plan to rebuild Alexa on a large language model, a wave of AI-generated Drake songs, and a leaked Google memo suggesting open source is outpacing Google and OpenAI. Read on for the full breakdown.

Transcript

welcome back everybody to AI news I do my best to try to do these every single Friday essentially this is your big stirred up Roundup of all of the AI news that you should try to keep track of all I'm saying viewers is if you are even slightly into AI if you're even slightly paying attention you're probably ahead of uh the majority of the populatio... Read More

Key Insights

• ✋ Deepfloyd AI offers impressive image generation capabilities, showcasing realistic and high-quality outputs.
• 📡 ChatGPT plugins introduce advanced browsing features, signaling the potential of AGI-like abilities in the future.
• 🌥️ Alexa's focus on developing a large language model reflects the trend of enhancing virtual assistants with AI technology.
• 🎼 The growing popularity of AI-generated music indicates the significant impact AI is having on the music industry.
• ❓ ChatGPT's Code Interpreter demonstrates the potential of AI for data analysis, visualization, and other complex tasks.
• 🛟 OpenAI's GPT-4 with browsing and Chegg's AI product "Checkmate" show the increasing competition among AI models in serving niche applications.
• 🤗 Open source AI technology poses a challenge to established companies like Google and OpenAI, highlighting the advantages of transparency and accessibility.
• 🤨 The departure of the "Godfather of AI" from Google raises questions about the risks and implications of rapid AI development.
• 👊 Bing AI updates and advancements in voice cloning, multi-modal chat, and image generation illustrate the continuous progress in AI technology.

Questions & Answers

Q: What is DeepFloyd IF and why does it matter?

DeepFloyd IF is a fully open-source AI image generator with upscaling and inpainting built in, and it is known for very realistic generations and for spelling text better than any model the creator had seen. It produces a zero-shot FID-30K score of 6.66, beating DALL-E 2, Parti, and Stable Diffusion on benchmarks. Being open source is the key part, since the community can iterate on it the way it did with Stable Diffusion.

Q: What browsing capabilities are coming to ChatGPT?

ChatGPT is getting an official OpenAI browsing model that can search the web, click a specific link, read the context, and then search again, similar to the small AGI-style agents. Access is very limited for now. Browsing is also coming to GPT-4 mode, not just default GPT-3.5, which the creator notes is far better.

Q: What can ChatGPT's Code Interpreter do?

The Code Interpreter can perform data analysis, visualization, business strategy work, academic research, paper writing, and basic video editing. It can generate accurate graphs, run crime hotspot analysis, and create hypotheses from datasets. The creator describes it as truly mind-blowing.

Q: How is Amazon planning to change Alexa?

Amazon's CEO, Andy Jassy, shared plans to revolutionize Alexa by developing a cutting-edge large language model. The goal is to make Alexa the ultimate AI personal assistant, capable of the complex tasks it currently cannot handle. Amazon is also connected with Stability AI, which may factor into the project.

Q: How is AI impacting the music industry?

AI-generated music is already taking off, highlighted by AI Drake songs such as "Winter's Cold" and a full AI Drake album the creator recommends. Beyond Drake, there is a list of roughly a hundred AI hit songs using popular artists' voices, including Travis Scott, Kanye West, and Lil Uzi Vert. The creator frames this as the early days of infinite AI-generated music.

Q: How is open source AI affecting Google and OpenAI?

Leaked material from Google suggests open source AI models are advancing rapidly and outcompeting both Google and OpenAI. Open source options are faster, more customizable, and more cost-effective, which pressures the big labs to prioritize third-party integrations and unique value adds. The episode frames this as a real challenge to established players.

Q: Who is the "Godfather of AI" and what did he do?

Jeffrey Hinton, often called the "Godfather of AI," recently left Google and warned about the potential dangers of AI. His departure raises questions about the risks and implications of how fast AI is developing. It underscores the need for careful consideration as the technology advances.

Q: What is Chegg's "Checkmate" AI product?

Chegg plans to compete with ChatGPT by releasing its own AI product called Checkmate. It was developed in collaboration with OpenAI and is aimed at serving Chegg's specific customer base with personalized educational support. It is an example of companies building niche AI tools on top of existing models.

Summary & Key Takeaways

• Deepfloyd AI, an open-source AI image generator, offers realistic image generations with built-in upscaling and inpainting capabilities.

• ChatGPT plugins introduce browsing functionalities to search and click on links, allowing for advanced browsing similar to AGI agents.

• Alexa plans to incorporate a large language model to enhance its capabilities as an AI personal assistant.

• AI-generated music is gaining popularity, with artists like Drake creating AI songs, and the music industry is embracing AI-generated hits.