• D.C.
  • BXL
  • Lagos
  • Riyadh
  • Beijing
  • SG
  • D.C.
  • BXL
  • Lagos
Semafor Logo
  • Riyadh
  • Beijing
  • SG


In today’s edition, we bring you highlights from Semafor’s Next 3 Billion Summit about connecting th͏‌  ͏‌  ͏‌  ͏‌  ͏‌  ͏‌ 
 
rotating globe
September 25, 2024
semafor

Technology

technology
Sign up for our free newsletters
 
Reed Albergotti
Reed Albergotti

Hi from New York, where the UN General Assembly is in full force and, of course, AI is a big topic as nations gather to discuss the biggest issues on the planet.

At Semafor’s Next 3 Billion Summit Tuesday, world leaders and top executives joined us to discuss the effort to bring connectivity to the almost third of the global population that isn’t currently connected.

That endeavor has been long in the making. Increasingly, it also means bringing AI to emerging markets. The internet boom created unprecedented wealth in the US and Europe, but in many ways, failed to lift up the rest of the world. And there’s a sense, or worry, that AI will follow the same plot line.

Below, read some of the highlights from the event.

Move Fast/Break Things
Toby Melville/Reuters

➚ MOVE FAST: Joining. Film director James Cameron is the newest board member of Stability AI, the embattled company known for the Stable Diffusion image creator. The big name is a vote of confidence for a firm whose future was in doubt and the move may signal a turnaround.

➘ BREAK THINGS: Severing. When Elon Musk fired most of Twitter’s employees, he refused to pay severance, spurring workers to take him to court. He recently lost an arbitration, forcing him to pay one former employee’s severance package, Bloomberg reported. The decision could encourage more ex-staffers to sue.

PostEmail
Artificial Flavor
Courtesy of Digital Green

Meta has spent billions of dollars on its Llama family of AI models, which it gives away for free under an open source license. There’s a strategy behind this, but it has also put Meta in the position of being the leading advocate for open source as the industry’s preferred method to develop AI (some AI safety advocates disagree). As part of that strategy, Meta on Tuesday announced the recipients of the Llama Impact Grants, which funds nonprofit organizations doing public-interest work with Llama models.

One of the projects, called Farmer Chat and made by Digital Green and One Acre Fund, fine tunes an AI model to give farmers in the developing world up-to-date advice on things like optimal planting times. Open-source models can be highly customized and run as efficiently as possible. I asked the Farmer Chat creators how much this advice would cost to provide for free to farmers in Africa. Their response: They want to get it down to 35 cents. That’s compared to $30, they said, without using AI. That really highlighted to me how open-source models are likely the way forward in much of the world.

PostEmail
The Next 3 Billion
PostEmail
Semafor Stat
700,000

The number of Ray-Ban Meta glasses the company has shipped since it launched the second generation in 2023, according to this Bloomberg profile of the woman who leads the division. The glasses have been a surprise niche hit, despite historic stigma around face-worn devices.

PostEmail
Obsessions
Amir Cohen/Reuters

OpenAI CEO Sam Altman’s blog post, The Intelligence Age, was published Monday and created a big stir in the tech industry. Not all of it was positive, and a lot of that has to do with Altman’s unapologetically positive view of AI’s impact on humanity without sufficiently examining the risks. Altman didn’t say anything much different from what he’s said before. Here’s my breakdown:

  1. OpenAI might achieve superintelligence in as little as 8 years. (Not a surprising belief, based on his past comments.)
  2. Sooner than that, we’ll have AI agents that can act autonomously, like a human assistant might do work for us. (Everyone is working on this.)
  3. AI will lead to abundance. The poorest person in the future will live better than the richest person alive today. (He’s said the same thing before and while it sounds out there, it’s been true throughout history. I’d rather be alive today than be a king in Medieval Europe, with high rates of infant mortality, disease and short life spans.)
  4. If we continue to build bigger data centers to run larger and larger AI models, we will eventually get to superintelligence. (This may be the most important point in the piece. Not everybody agrees with this and even Altman has contradicted this concept in the past. It could mean OpenAI has seen promising results from larger compute clusters.)
  5. At some point, AI will be able to improve itself, essentially taking humans out of the process of building better AI models, altogether. (Not a new idea, but interesting that he says it with such confidence.)
  6. People might fight wars over AI, unless it becomes an abundant resource. The way to do that is to build so much infrastructure that AI becomes inexpensive. (This one should be unpacked. Unclear if he’s making a point about the AI race between the China and the US.
  7. We should focus on the downsides of AI now. In particular, economic ones. (Only notable because he didn’t mention existential risk.)
  8. People will always have jobs, but we can’t imagine jobs in an AI age for the same reason a Medieval farmer would not have predicted that social media influencers would one day exist. (Fairly common view for technologists.)
PostEmail
Hot on Semafor
  • The data center boom is giving clean energy a jolt.
  • Democrats fret as GOP super PACs pour money into Senate battle.
  • Wall Street jockeys for jobs as election nears.
PostEmail