📢 ChatGPT is listening

Plus Getty gets into AI

Thumbnail showing the Logo and a Screenshot of 📢 ChatGPT is listening

Chatting with ChatGPT

Following last week’s quality-of-life improvements for Dall-EOpenAI has added a fresh coat of paint to ChatGPT.

The AI company is rolling out an update that will let users create prompts by speaking to ChatGPT, or by simply uploading a picture. To use your voice, you just tap a button and state your request. Image support is a little more interesting, as you can ask the chatbot questions about a picture, and even dig into details across multiple images.

OpenAI is definitely playing catch up with Google’s Bard here, but it’s great to see voice and image prompts become table stakes in the world of generative AI bots.

OpenAI on X | Link to story.

If you can’t beat ‘em, join ‘em

Getty Images, the photo licensing operator you know from all those watermarked images, is jumping into the AI game.

The company has partnered with Nvidia to launch Generative AI by Getty Images (GAIGI). As the poorly-thought-out name suggests, it’s an AI-powered image generation tool trained solely on Getty’s enormous library of photos. And the most interesting part? Getty says it’ll pay photographers if their content was used to inform an AI image.

It sounds like a compelling service - particularly if GAIGI can generate legitimate photo-realistic results. But perhaps the best part is that Getty promises there will be no watermarks on the final graphic. GAIGI is rolling out via Getty’s website.

The Verge | Link to story

Amazon delivers billions of dollars to Anthropic

Amazon has announced it is investing a cool $4 billion in AI company Anthropic.

As part of the arrangement, Anthropic will use AWS for its online computing needs. It will also work with Amazon on generative AI projects.

Anthropic, which is perhaps best known for its Claude LLM, was established by some former OpenAI executives. It had previously raised a solid $1.4 billion.

CNBC | Link to story

The intern is a robot

The conversation about AI supplanting workers is not a new one, but what if people are inadvertently training the very machines that will ultimately replace them? That’s the premise of a new think-piece from Wired, in which the publication explores the prospect of companies using subversive training software in the workplace.

MIT economics professor David Autor suggests tools can learn from humans as they’re being used. Autor also posits the endgame may not even be machines taking the place of humans at work, but rather that certain jobs will simply cease to exist over time.

Wired | Link to story (paywall)

Profile Picture of Tom Wilton

Written By: Tom Wilton

Lead Newsletter Writer

Published Date: Sep 26, 2023

Explore OpenAI's new ChatGPT features, Getty's AI imagery, Amazon's hefty investment in Anthropic, and AI's workplace impact.