- Starter AI
- Posts
- All about: Build 2024, Meta’s Chameleon, Big tech pledges AI safety
All about: Build 2024, Meta’s Chameleon, Big tech pledges AI safety
More committed than ever.
Sthefania
May 22, 2024
In partnership with
Hello, Starters!
AI started roaming as freely as possible, but as it grows and enters more industries, countries and big tech companies are coming together to lay the groundwork for comprehensive regulations.
Here’s what you’ll find today:
Microsoft’s announcements at Build 2024
Meta presents its Chameleon model
Tech giants commit to AI safety
Adobe’s “Generative Remove” lands on Lightroom
xAI plans to turn Grok multimodal
And more.
📣 Microsoft’s announcements at Build 2024 (2 min)
Unexpectedly, many of the announcements at Microsoft's Build 2024 conference involved AI. The company is introducing products and collaborations that solidify its spot in the AI race, including a partnership with Khan Academy for an AI-driven teaching assistant, Team Copilot for meetings, and a new Phi-3 multimodal model.
Team Copilot is the new extension of Copilot that turns it into a main asset for your meetings, as it can manage the agenda, take notes, keep track of time, and even lead the meeting. On the other hand, Phi-3-Vision is the latest addition to the Phi-3 family. It can process both images and text and is optimised for mobile devices.
🦎 Meta presents its Chameleon model (3 min)
Multimodal is the way to go for most of the recent models. Meta's upcoming Chameleon is one of them. Powered by an "early-fusion token-based mixed-modal" architecture, Chameleon has been fundamentally designed as multimodal; it doesn't require other specific components to do its job.
It has learned from a mix of images, text, and code and processes images similarly to how language models handle words, by turning them into tokens. Meta has yet to release the model.
🤝 Tech giants commit to AI safety (1 min)
Putting safety first, a group of companies, including major names like OpenAI, Microsoft, Amazon, and others from different countries, has come together at the Seoul AI Safety Summit to pledge frameworks that ensure the responsible development of AI technologies, preventing their misuse by bad actors.
The framework includes intolerable risks, such as the threat of bioweapons and cyberattacks, in which companies will stop the development of their models if they can't ensure these risks are mitigated. This is one of many efforts happening worldwide as more countries realise the need for regulations in the field.
Have you heard of Prompts Daily newsletter? I recently came across it and absolutely love it.
AI news, insights, tools and workflows. If you want to keep up with the business of AI, you need to be subscribed to the newsletter (it’s free).
Read by executives from industry-leading companies like Google, Hubspot, Meta, and more.
Want to receive daily intel on the latest in business/AI?
🎨Adobe continues to bring AI to its offerings through Firefly, and this time, Lightroom is the chosen one with the integration of "Generative Remove," an AI-powered feature that allows creatives to easily edit out objects from their photos, no matter how complex they may be.
🤖Recent developer documents have revealed that xAI is continuously working on making Grok multimodal, an endeavour teased last month. Users might soon be able to upload photos to the chatbot and receive responses based on them.
⚡️Quick links
What did you think of today's newsletter? |