Google Artificial Intelligence AI

Weekly AI Recap: Meta launches Llama 3.1 & OpenAI unveils online search tool

Author

By Webb Wright, NY Reporter

July 26, 2024 | 6 min read

Also, former NFLer and social activist Colin Kaepernick has debuted his own AI company.

Llama 3.1

Meta called Llama 3.1 405B “the world’s largest and most capable openly available foundation model.” / Meta

Meta releases Llama 3.1 405B

On Tuesday, Meta announced the public release of Llama 3.1 405B (denoting 405bn parameters), described in a company blog post as “the world’s largest and most capable openly available foundation model.” The company also introduced Llama 3.1 70B and 8B, updated versions of open models that were released earlier this year.

Llama 3.1 405B is designed to compete with GPT-4o and Claude 3 Sonnet; Meta claims the model offers ”state-of-the-art capabilities in general knowledge, steerability, math, tool use and multilingual translation.” The product is also being positioned as a cutting-edge tool for the generation of synthetic data, which can, in turn, be used to train smaller language models.

Meta CEO Mark Zuckerberg took the launch as an opportunity to double down on his belief that open-sourcing is the future of AI, saying in a blog post that it “represents the world’s best shot at harnessing this technology to create the greatest economic opportunity and security for everyone.”

Powered by AI

Explore frequently asked questions

OpenAI introduces SearchGPT

OpenAI is officially going head-to-head with Google. On Thursday, the ChatGPT parent said it has released a prototype of SearchGPT, a new online search tool powered by a large language model.

The tool – which is currently in beta and being tested by a handful of news publishers – is designed to deliver up-to-date information from the web via conversational AI à la ChatGPT.

OpenAI is positioning SearchGPT as a boon for publishers. According to the blog post, the search tool will include citations and links to relevant sources, not unlike those used in the AI-powered search platform Perplexity.

Colin Kaepernick unveils AI startup

Colin Kapernick, the former NFL quarterback who stirred controversy in 2016 with his on-field political protests, launched a new AI startup on Wednesday.

The company, Lumi Story AI, is intended to provide creators with AI-powered resources that can help them overcome certain creative or commercial hurdles. It has raised $7m in funding, led by Reddit cofounder Alexis Ohanion’s VC firm Seven Seven Six. Other notable investors include Minted CEO Mariam Naficy and the rapper Chamillionaire.

Lumi is available now in beta for comic book and graphic novel creators, though it plans to eventually expand its suite of product offerings, according to a report from The Wall Street Journal.

Etoro Olympics ad highlights shortcomings of AI-generated video

On Wednesday, financial services company Etoro released a 30-second video spot created using an AI-powered video generation tool called LTX Studio. The ad will run during the Paris Olympics, which begins today.

Like almost all AI-generated videos, the new video spot includes some strange visual quirks; in one scene, for example, a man – intended to represent a fictitious Etoro investor – walks alongside a row of orange flowers, some of which are taller than he is, making the video look less like an ad and more like a trippy music video.

Suggested newsletters for you

Daily Briefing

Daily

Catch up on the most important stories of the day, curated by our editorial team.

Ads of the Week

Wednesday

See the best ads of the last week - all in one place.

The Drum Insider

Once a month

Learn how to pitch to our editors and get published on The Drum.

Google is the only search engine that will surface Reddit content, report finds

Google is now the only major search engine that will yield results from Reddit, according to a report published Wednesday by 404 Media. Most other search engines, including Bing and DuckDuckGo, have reportedly been blocked from crawling Reddit data.

Reddit denied the report’s claim that Google’s apparent hegemony over surfacing results from Reddit is part of a $60m deal between the companies, which, signed five months ago, gives Google the right to scrape content from Reddit in order to train AI models. Rather, the company said that it has been blocking all crawlers that have not committed to refraining from using data scraped from the site in order to train AI models.

For more on the latest happenings in AI, web3 and other cutting-edge technologies, sign up for The Emerging Tech Briefing newsletter.

Google Artificial Intelligence AI

More from Google

View all

Trending

Industry insights

View all
Add your own content +