Is Meta making AI capabilities more flexible and finding new business opportunities?

This is the published version forbes magazine CIO Newsletter. Provides the latest news for chief innovation officers and other technology-focused leaders. Click here to get it delivered to your inbox every Thursday.

---

Meta has had a great week, but you wouldn't think so by looking at the stock chart. The social media and programming company's stock fell more than 11%. Wednesday's quarterly results beat expectations, posting 27% year-over-year sales growth. All of this comes less than a week after the company announced what it calls an innovative AI chatbot with advanced open LLM.

Meta's chatbot is highly praised not only for its functionality, but also for its accessibility.. The chatbot can provide answers, collate information from the internet, create images from text prompts, and search Meta's social media feeds, performing the same types of tasks as ChatGPT and Google's Gemini. However, Meta's is integrated with popular social media platforms and rather than redirecting users to another his website, it is placed directly where users are already located. (Meta's chatbot is also available through its own URL.)

but Provides more opportunities for developers,write forbes Senior Contributor Janakiram MSV. Llama 3 LLM is much more advanced than other models and outperforms several other models in various parameters such as writing and coding in English. It is now available to developers and is already integrated into several developer ecosystems, including Hugging Face. Currently, it can support a context length of 8,000 tokens. This means you can respond to longer queries with more detail. And Meta continues to develop his Llama 3, aiming to expand its features, language, and versatility. The company expects new products to launch later this year.

After rolling out a new AI chatbot and LLM and posting nearly $36.5 billion in quarterly revenue, the drop in Meta's stock price seems shocking. but, The company expects ad sales, which had been growing rapidly, to start slowing., expects second-quarter sales to be lower than the average analyst estimate. The company also raised its full-year spending forecast.

CEO Mark Zuckerberg spoke at an earnings conference. Meta is currently working on building and scaling AIis investing in development and not yet focused on monetization. But Meta has been here before, he said. “Historically, investing in building new scale experiences like this in our apps has been a very good long-term investment for us and for the investors who stick with us. “I did,” he said.

Investors and people working in the technology industry are two very different groups with different priorities..However, as forbes Contributor Alex Zhavoronkov writes that Llama 3 changes everything in the technology field. That LLM will likely be what developers choose to build, and just like in the social media space, meta will prevail in this space as well. Meta's social media ad revenue growth may slow, but the coming quarters will show how the company takes big steps in AI and the opportunities there.

Over the years, there has been a lot of discussion about open source code, software, and LLMs. Historically, open data did not exist, but that is changing. Marc Prioleault, executive director of the Open Maps Foundation, told me: Why open data is important for the future. Excerpts from our conversation will be published later in this newsletter.

Policy + Regulation

After years of talking about banning TikTok, the federal government actually banned it this week.. A measure that would force Chinese company ByteDance's short video app to be sold to U.S.-friendly owners or eventually be banned from app stores joins a series of bills providing military aid to Ukraine, Israel and Taiwan. added. The bill passed both houses of Congress and was signed by President Joe Biden yesterday.

And now the legal battle begins. forbes magazine Report by Alexandra S. Levine and Emily Baker-White TikTok isn't going down without a fight. A weekend internal memo from Michael Beckerman, TikTok's head of Americas policy, promises a legal battle. “This bill is a clear First Amendment violation of the rights of the 170 million Americans who use TikTok,” he wrote. Many say that selling TikTok is also not a realistic goal. forbes Many of TikTok's internal systems are built on ByteDance tools custom-built by Chinese engineers several years ago and are difficult to dismantle, according to reports. Former National Security Agency general counsel Glenn Gerstel said it was TikTok's advanced algorithms that made a sale “economically impractical” and that China would be reluctant to sell it.

but, forbes Contributor Peter Suchiu points out that TikTok's problems may begin Another much-needed conversation among policymakers: protecting data privacy. The law targets TikTok because China used the app's data to spy on Americans. But in the US, data from apps can also be used in invasive ways. Approov CEO Ted Miracco told Suciu that the bill is a “blunt weapon” that doesn't address the root causes of the concerns, and that legislation that enacts strong privacy protections for all apps, regardless of their home country, would be much better. He said it could work effectively.

Technology + Innovation

Apple's big push into VR headsets may not be paying off. Citing a report by Ming-Chi Kuo, an analyst at TF International Securities, forbes Senior contributor Paul Tassi writes that Apple will roughly cut the number of Vision Pro VR units it ships to the US this year by half, from 700,000 to 800,000 and from 400,000 to 450,000. Tassi says this is unlikely to be due to the price of the device, although it is an expensive purchase at $3,500. More fundamentally, people don't want VR headsets. This is a lesson Meta has learned in recent years, as its VR arm, Reality Lab, has consistently posted losses. In yesterday's earnings report, the Meta division reported a loss of $3.8 billion, and the company expects year-over-year losses to be “significantly higher.”

artificial intelligence

The technology underlying AI is interesting, but Stanford University's Institute for Human-Centered AI The new annual AI Index is a reminder that it is a work in progress.. forbes Senior contributor Joe McKendrick says that while AI can now beat humans at some tasks, there are still complex areas where humans can beat them, such as competitive-level mathematics, visual common-sense reasoning, and planning. It says that there are some. And AI developers themselves have “low transparency scores,” hampering efforts to see how robust and secure their AI systems are. It also counts the number of “malicious AI” incidents, which increased by 32.3% last year.

bits + bytes

How mapping projects point the way to more open data usage

The Overture Maps Foundation, a project to standardize map data, was founded in December 2022 by Amazon, Meta, Microsoft, and TomTom and hosted by the Linux Foundation. I spoke to Marc Prioleau about the projects he's working on that rely on open data from a variety of sources, and about the future of open data. This conversation has been edited for length, clarity, and continuity.A longer version of this interview is available here.

Why do you think open data is so new?

I think there are many similarities between open code and open data, but there are some important differences. One of the similarities between open code and open data is that many people agree that building things together produces better results and higher quality products. I think it's true. Typically with open code and open data, there is an “open” part, but people still have the ability to create business models on top of it. I think it took Open Code a while to figure it out, but it's pretty well established now. Open data is experiencing the same thing. What are the areas where data can be shared and open, but still leave room for business models on top of it?

There are some differences. Let's say you're a software engineer. When you write code in an open source project, it becomes open as soon as you post it. Data is a little different because data represents facts. It has to come from somewhere. It doesn't just come from your brain. It was a measurement of something. Cartography allows you to measure whether roads still exist. Is that business still there? Has things changed? What's upstream in open data and different from open code is that there's data collection, synthesis, and analysis, but it usually doesn't necessarily have to be open. I think this is a very different business proposition than open code.

What do you think needs to happen to get more companies and other organizations interested in open data?

My feeling is that conceptually everyone understands the strategy around open data. They get it, nod and do everything. The challenge is to get the group together to actually collaborate. One of the challenges in open source projects is how to share participation. Are there so-called free riders, people who use the work of an open source project but don't contribute to it? After all, if no one participates, the project dies. I think the important thing is to develop mechanisms where people can contribute and have an incentive to contribute.

Strategically, at least for our project, I can explain it to people and they understand the strategy, the rationale, and why they want to do it. Why it makes sense. I don't think that's the difficult part at all. But you know, in that case people have to do it. They have to go far. We started with 4 members. There are currently 27 people. Part of it is the benefit they get if this happens and the alternatives if it doesn't happen. The alternative is not good.

In five years, will open data be viewed the same way as open code and open software are today?

I think they are related, but separate fields. Today, people think of open source as open source. Code data is irrelevant. Five years from now, I think open data will be a separate entity. It's because of these things we talked about. Data has to come from somewhere, so you need people to provide something proprietary and valuable. Data must be accurate. To output data accurately and quickly, it must be output at a specific frequency. The other thing is that the data is large. We're dealing with terabytes and petabytes of data. This will incur the cost of cloud computing.

I think companies will start to develop expertise around that. I think open data infrastructures like ours have to deal with these unique things in unique ways.

facts and comments

The legal dispute between Apple and Epic Games is back in court The next month, a judge ruled that Apple's fees for third-party purchases appear to ignore the court's requirement that users create alternative ways to make purchases.

27%: Apple charges fees for purchases from large developers outside the App Store

3% to 6%: Amount Epic CEO To tell Non-App Store payments involve third-party payment processor fees, making them more costly for developers

“Undermining the spirit of the injunction”: District Judge Yvonne Gonzalez Rogers wrote that Apple's actions appear to restrict competition, impede the free flow of information, and constrain user choice.

Strategy + Advice

Managers are not always loved by their team members. Look in the mirror and make sure there's nothing wrong with you. Here, we will introduce how to identify bad management yourself and give tips for improving it.

There are many things that leaders do. remain unseen and unnoticed. Here we explain how to perform some of these tasks.

quiz

Axon is best known for manufacturing Tasers. New AI-powered tools for law enforcement. What is it for?

A. Automatically track suspicious vehicles using traffic camera data

B. Create a police report from body camera audio

C. Cross-referencing records to identify repeat offenders

D. Flag when equipment is ready for maintenance

Please check if your answer is correct here.