AI Roundup 019: The church of ChatGPT

June 16, 2023

Artwork created with Midjourney.

The church of ChatGPT

This week a University of Vienna philosopher used ChatGPT to create a 40-minute church service. It included a sermon, prayers, and music, and was led by four AI-generated avatars.

Why it matters:

• AI is eating the world, and spirituality and religion are not immune. This service was just an experiment, but there will be real zealots in the future.

• Right now, education is at the forefront of AI’s real-world impacts. The New York Times examines how schools are scrambling as "AI literacy" becomes the new education buzzword.

• We've also seen plenty of AI content upend creative writing and photography. But there's a surprising new market dealing with a generative AI flood: Etsy, the place for handmade crafts.

Artwork created with Midjourney.

AI Geopolitics

The US and China have had a rocky relationship for several years. Technology is a key part of the conflict, and that looks to be expanding to AI.

Between the lines:

• Microsoft is reportedly moving AI researchers from Beijing to Vancouver to avoid losing them to Chinese companies.

• Tech companies have withheld their chatbots from China for geopolitical reasons, and it appears Hong Kong is included in that decision too.

• Sam Altman, still on his world tour, called on China to help shape AI safety amid global cooperation.

Elsewhere in AI geopolitics:

• The EU passed a draft of its AI Act, which would bring heavy regulation to generative AI. A final version is expected to pass later this year.

• UK Prime Minister Rishi Sunak wants the UK to become an AI research and safety hub. He's starting with DeepMind, OpenAI, and Anthropic opening up their models to the UK government. He’s also planning an AI summit for this fall.

• Here in the US, Republican Senators want to exclude generative AI from Section 230 protections. Section 230 is the law that stops Big Tech from being sued over user-uploaded content, and some experts argue it should cover generative AI.

Artwork created with Midjourney.

Meta's many models

Meta continues its march to capture the open-source movement. It's impressive how little competition it faces from other tech giants around open-source AI.

What to watch:

• The big news was MusicGen, an AI model that can generate music clips from text prompts and existing melodies.

• The company also detailed I-JEPA, a computer vision model to create more accurate images.

• It's reportedly planning to release a commercially-licensed version of LLaMA, the leaked research model that now powers hundreds of open-source projects.

Elsewhere in the FAANG free-for-all:

• The WSJ looks at Microsoft and OpenAI's unusual partnership and the internal conflicts it created.

• Google Shopping adds a virtual "try on clothing" feature to preview items on AI models, and Google Lens can help identify skin conditions.

• Amazon is testing generative AI to summarize product reviews and surface the most liked features.

Things happen

Paul McCartney is working on “the final Beatles record” with AI vocals from John Lennon. OpenAI adds to ChatGPT’s code-execution capabilities. LUUM, the studio using robots to apply eyelash extensions. Nvidia’s AI Red Team. Stable Diffusion’s bias problem. Mistral AI raises $113M despite not having a product. Mayoral candidate fakes campaign imagery with AI. When your nurses get overruled by AI. Alphabet warns staff about the dangers of using its chatbot.

Last week’s roundup