The week in AI: ChatGPT news and upgrades

Welcome to The Dispatch! We are the newsletter that keeps you informed about AI. Each Thursday, we aggregate the major developments in artificial intelligence; we pass along the news, useful resources, tools or services, and exciting projects in open source. Even if you aren’t an engineer, we’ll keep you in touch with what’s going on in AI.

The web-browsing update is live now; the voice and image capabilities will be rolled out over the next two weeks. Voice is coming on iOS and Android (opt-in in your settings), and images will ultimately be available on all platforms including the free version of ChatGPT.

In our social media/videos section below (ChatGPT has been supercharged!), Two Minute Papers covers the upcoming multimodality. We share their enthusiasm - pretty incredible stuff from OpenAI.

Google Bard users' private conversations with the chatbot are being indexed by Google Search, potentially exposing sensitive information. SEO consultant Gagan Ghotra discovered links to Bard chats showing up in search results. Bard provides an option to share conversations (colleagues, friends), but most users likely aren’t aware this could make their chats suddenly become public.

After getting called out on social media, Google acknowledged this was ‘unintended’ and that they’re working to prevent Bard conversations from appearing in search results without users' consent. These repeated privacy issues reflect poorly on Google, who is currently in the midst of an anti-trust case.

Amazon is investing up to $4B into AI startup Anthropic as part of a broad collaboration to “develop the most reliable and high-performing foundation models in the industry”. AWS users will find Anthropic’s Claude LLM more deeply embedded through Amazon Bedrock, to include model customization and fine tuning.

This looks like a big win for both sides. Claude is building an increasingly impressive portfolio of enterprise users across industries, including LexisNexis, SourceGraph and thousands of other companies. If you haven’t given Claude a shot yet, you should - it’s excellent. Anthropic recently posted a helpful ‘prompt-engineering’ guide to help users make the most out of Claude’s impressive 100k token long context window (great for analyzing long documents/research - even books).

Spotify is launching a new pilot program that uses AI to provide translations of podcasts into different languages - all while retaining the original speaker's voice. ‘Voice Translations’ will leverage OpenAI's new voice generation capabilities to mimic the speaking style and vocal characteristics of podcast hosts as their shows are translated. Several top podcasters are participating in the initial trial, and voice-translated episodes from pilot creators will be available worldwide to both premium and free users. We listened to a Lex Fridman episode in Spanish - the quality of the translation and the voices (even down to the intonation) aren’t perfect, but still very compelling.

More AI this week:

• Hollywood writers will not be getting replaced by AI: Writer’s Guild strike officially over after 148 days

• OpenAI is reportedly raising funds at a valuation of $80 billion to $90 billion

• Microsoft is facing increasing scrutiny for its water usage in Iowa during drought-times

• The acronym behind our wildest AI dreams and nightmares

• Nature is launching a comprehensive new series on the emerging role of AI in science - both the good and the bad.

• Nvidia is partnering with Getty Images for a new risk-free AI-image generator

• 'The Creator' review: A stunning reminder we need more original sci-fi

That’s it for the week! See you next Thursday.