Here's who gets paid for AI

PLUS: US-EU differ on AI policy, Open AI's custom chips and Anthropic's new research.

Sign up | Advertise | Ben’s Bites News
Daily Digest #257

Hello folks, here’s what we have today;

PICKS

  1. Meta is paying celebrities millions of dollars to license their likenesses for AI chatbot characters. Reminder: These celebrities are playing a personalized avatar - Charli D'Amelio's is a dancer named Coco, while MrBeast's AI is big bro Zach on Meta apps that people can chat with. 🍿Our Summary (also below)

  2. US warns EU’s landmark AI policy will only benefit Big Tech. The US is concerned some provisions are vague and that the EU focuses too much on how AI models are developed rather than their risks.🍿Our Summary (also below)

  3. OpenAI is exploring making its own AI chips to solve its shortage of expensive AI chips that power its software and products. The search for new options has been going on since last year. Currently, they rely heavily on Nvidia's graphics processing units (GPUs) and Microsoft’s partnership to access them. 🍿Our Summary (also below)

  4. Decomposing language models with dictionary learning. Anthropic’s new research extracts a large number of interpretable features from a one-layer transformer. 🍿Easy to read summary.

TOP TOOLS
NEWS
QUICK BITES

Meta is paying celebrities millions of dollars to license their likenesses for AI chatbot characters. Reminder: These celebrities are playing a personalized avatar - Charli D'Amelio's is a dancer named Coco, while MrBeast's AI is named Zach on Meta apps that people can chat with.

What is going on here?

Meta is shelling out millions to get celebrities to license their likenesses for AI characters in a bid to draw users to its platforms.

What does this mean?

According to someone with inside knowledge, Meta is paying stars like Snoop Dogg, Tom Brady, MrBeast and Charli D'Amelio millions of dollars over multiple years to use their likenesses for fictional AI avatars. For example, Meta is paying one top creator around $5 million over two years for about six hours of studio work. The deal is also non-exclusive, which means it allows the celebrities to also work with other AI services.

The AI avatars are currently text-based chatbots on Meta's platforms, but Zuckerberg said voice capabilities are coming in 2023.

Why should I care?

Meta is still all-in on its vision for the metaverse and AI, despite its recent struggles. Meta seems willing to pay top dollar to partner with big names who can draw their massive audiences to use the AI avatars. If the celebrity avatars take off, they could be a blueprint for how creators monetize their brands in virtual worlds. There's also a chance Meta pulls the plug on funding if user traction is low, just as it did with Facebook Watch originals.

QUICK BITES

Open AI is exploring making its own AI chips to solve its shortage of expensive AI chips that power its software and products. The company relies heavily on Nvidia's graphics processing units (GPUs) and Microsoft’s partnership to access them.

What does this mean?

Open AI sees in-house chip development as key to tackling two big problems: the global scarcity of high-end AI chips and the massive computing costs of running systems like ChatGPT. CEO Sam Altman has made acquiring more AI chips a top priority given Open AI's reliance on Nvidia, which controls over 80% of the AI chip market.

Why should I care?

Custom AI chips could significantly reduce Open AI's hardware costs and chip shortage issues, allowing it to scale its AI technologies more easily. However, building custom AI chips is a complex and expensive endeavour that could cost hundreds of millions of dollars with no guarantee of success. Even if Open AI acquires a chip company, it would still take years to develop its own chips.

QUICK BITES

The folks at US State Department have done an analysis focused on the EU Parliament's version of the AI Act, which includes rules on generative AI. They think that this AI Act will only benefit Big Tech.

What is going on here?

The US warned the EU that its AI Act may disadvantage smaller firms and hurt EU competitiveness.

What does this mean?

The US is concerned some provisions are vague and that the EU focuses too much on how AI models are developed rather than their risks. This could dampen productivity gains and investment in the EU, limiting competitiveness. The rules may also hamper investment in AI research and commercialization in the EU, since training large language models requires substantial resources.

The US shared this feedback with EU counterparts in recent weeks. Some of the concerns echo those of EU member states regarding the Parliament's proposal. This analysis aligns with the US preference for a light touch approach to AI regulation to foster innovation. But the US has sent mixed messages, with some officials warming to the idea of mandatory rules amid ethical concerns about AI.

Why should I care?

The US analysis highlights key differences in regulatory approaches between the US and EU. How the EU regulates foundational AI technologies could impact innovation and the competitiveness of EU firms. If regulations are too stringent, investment and jobs may migrate out of the EU. But lax rules also risk potential harms from uncontrolled AI advances. Finding the right balance will shape the future AI landscape.

Ben’s Bites Insights

We have 2 databases that are updated daily which you can access by sharing Ben’s Bites using the link below;

  • All 10k+ links we’ve covered, easily filterable (1 referral)

  • 6k+ AI company funding rounds from Jan 2022, including investors, amounts, stage etc (3 referrals)

Reply

or to participate.