Daily AI Podcast

Introduction

Happy Saturday everybody! It’s April 29, 2023 and I’m Dean from the Daily AI Podcast.

In today’s headlines, we have news about OpenAI’s funding and business plan as well as
Amazon developing a more generalized and capable LLM for Alexa.

Our Policy and Ethics segment will cover developments in the EU policy and a legal case that could have large implications in copyright law and AI. In our learnings segment, we’ll be highlighting a new course for ChatGPT Prompt Engineering.

Finally, today’s daily feature will be of great use for anyone who wants to find useful AI tools
but doesn’t know where to start. So sit back, relax, and let’s dive into today’s stories.

INDUSTRY NEWS

OpenAI closes $300M share sale at $27B-29B valuation
Source

OpenAI, the creators of the widely-used AI language model ChatGPT, has raised over $300M in a tender offer that values the startup at $27B-$29B. New investors include Tiger Global, Sequoia Capital, Andreessen Horowitz, Thrive, K2 Global, and Founders Fund. This investment is separate from the $10B investment by Microsoft, which was completed earlier this year. OpenAI now has over 30% of outside investors, including startups and large businesses, with businesses such as Google attempting to create their own generative AI services. Despite concerns about the security and ethics behind OpenAI’s services, the rapid development of its ecosystem demonstrates great interest in AI with generative services at the forefront of investment priorities.

Amazon is building a ‘more generalized and capable’ LLM to power Alexa, says CEO
Source

According to CEO Andy Jassy, Amazon is in the process of creating a “more generalized and capable” large language model (LLM) to enhance Alexa’s capabilities. AWS has been working on large linguistic models for several years, as well as customized machine learning chips specifically for training called ‘Trainium’ and a prediction-based chip called ‘Inferentia.’ Jassy also discussed how companies will soon have the ability to customize foundational models through Amazon’s new managed foundational model service, Bedrock.

OpenAI previews business plan for ChatGPT, launches new privacy controls
Source

OpenAI is planning to introduce a new subscription tier for its ChatGPT bot for enterprise customers. Called ChatGPT Business, it is designed for professionals who need more control over their data and end users. The subscription will include a data usage policy that means that end users’ data won’t be used to train OpenAI’s models by default. According to OpenAI, the subscription service will be available in coming months. With more than 100 million monthly active users two months after launch, ChatGPT is said to be the fastest-growing consumer application in history. Operating expenses are “eye-watering,” with compute costs amounting to a few cents per chat, says openAI CEO Sam Altman. In a blog post, OpenAI announced today that all ChatGPT users will also now be able to turn off chat history, with conversations disabled when chat history is turned off not being used to improve OpenAI’s models.

POLICY AND ETHICS

EU lawmakers pass draft of AI Act, includes copyright rules for generative AI
Source

The European Union lawmakers have approved a draft of its Artificial Intelligence (AI) Act, marking the first comprehensive bill related to regulating AI. The draft will next be discussed during the trilogue, where EU lawmakers and member states will negotiate the final details of the bill. The AI Act requires generative AI models, which include tools like ChatGPT, to be developed under EU law and subject to fundamental rights, including freedom of expression. Companies that make generative AI tools would have to disclose if copyrighted material was used. Furthermore, the AI Act applies to companies providing or using AI systems in the EU, and those that are located third countries, such as the UK and the US.

The Andy Warhol Copyright Case That Could Transform Generative AI
Source

The US Copyright Office recently determined that AI-generated art is ineligible for copyright protection. Although artists can register works made with assistance from AI, they must display significant “human authorship.” Currently, a group of artists is suing makers of generative AI Midjourney, Stable Diffusion, and DeviantArt, claiming they are scraping artists’ works to train their models without permission. Several companies are filing motions to dismiss, saying AI-generated images bear little resemblance to the works they’re trained on. At the heart of these debates about AI’s impact on creative fields are questions of fair use. The US Supreme Court is expected to rule on Andy Warhol Foundation for the Visual Arts, Inc. v. Goldsmith, a case that will determine whether Warhol’s series of images of Prince were adequately transformative under the fair use doctrine of the Copyright Act. The ruling will determine when an act of creation begins and could greatly impact how copyright law is applied to what AI tools do with human-made works. AI is a keystone in global trade agreements, and granting AI copyright would fundamentally alter trade policy. The decision could further erode international relations.

LEARNINGS

Andrew Ng Collaborates with OpenAI for A New Course on ChatGPT Prompt Engineering
Source

Andrew Ng’s DeepLearning.AI and OpenAI have collaborated to launch a new course on ‘ChatGPT Prompt Engineering for Developers.’ The free course will teach learners how to use large language models (LLMs) to build powerful applications. Isabella Fulford from OpenAI and Andrew Ng will teach the course. The course will cover valuable tips for prompt engineering, demonstrate how LLM APIs can be used for summarization, inference, text transformation, expansion, and teach learners to create personalized chatbots. The course is beginner-friendly and accessible to novices with a fundamental grasp of Python, but also useful for experienced machine learning specialists. The course takes only 1.5 hours to complete.

Daily Feature

AI Finder

Source
Today’s Daily Feature is ai-finder.net, a website that can help you discover the best AI tools for any category you’re interested in. With so many AI tools being developed every day, it can be hard to keep up and know which ones are worth your time and investment. But with ai-finder.net, you can quickly and easily find the most highly recommended tools in any category, all in one place. I searched up ‘video editing’, ‘research assistant’, and ‘logo designer’ and was given relevant results that showed their capabilities, price, and rating. Whether you’re looking for an AI tool to help you with a specific problem, or want to see the most popular and newest AI tools, ai-finder.net can provide you with a list of the best options available. It’s a fantastic resource for anyone looking to take advantage of the latest innovations in AI.

Conclusion

And with that, we wrap up another episode of the Daily AI Podcast. We hope you’re feeling inspired to join the conversation about AI and help shape its future. After all, AI isn’t just a technology, it’s a community of innovators, thinkers, and doers. So keep sharing your ideas, keep exploring new possibilities, and keep listening to the Daily AI Podcast for all the latest updates.