Is Zuckerberg the New Jobs?

Today on the AI Daily Brief is Zuckerberg the New Jobs. The AI Daily Brief is a daily podcast and video about the most important news and discussions in AI. To join the conversation, follow the Discord link in our show. Notes. Welcome back to the AI Daily Brief. For the second time in two days, descript has lost an entire podcast file, so I actually won't be sharing the update about OpenAI's leadership shakeup. I'm already on my way to the airport so can't re record it and actually had to use 11 labs just to get this disclaimer. Sorry again and I'll provide the full update on OpenAI tomorrow. Today we are heading to 1 Hacker Way where Meta held its Connect Developer Conference and has unveiled a number of new things ranging from AI to ar. As you'll see, even in a day where we got a set of new models, it was really quite secondary to Zuckerberg's main presentation and I think was a reminder to some that as all in on AI as Zuckerberg and Meta have become, they have not left entirely the vision of the future which led them to rename the company Meta in the first place. First though, let's talk about the AI updates. TLDR is that Meta showed off the latest version of their open source models, llama 3. 2. There was llama 3. 2, 1B and 3B, which were designed specifically for on device use cases, but the more standard for developers were Llama 3.2, 11B and 90B. Vision models 11B and 90B were effectively multimodal alternatives to Llama 3.1, 8B and 70B, and indeed the multimodality is probably the biggest update. Meta said that the models quote support image reasoning use cases such as document level understanding, including charts and graphs, captioning of images, and visual grounding tasks such as directly pinpointing objects in images based on natural language descriptions. Two of the examples they gave a small business owner could feed llama 3.2 a chart of their last year of revenue and ask the model to highlight the months with the best sales. They could also show Llama a map and ask the model when a hike might become steeper or the distance of a winding trail. The models can be deployed with or without Meta's new safety tool called Llama Guard Vision, which can detect potential harmful texture images. The models are available to download from LLAMA directly as well as from Hugging Face, and they're also being integrated into Meta's wide range of cloud partners. As you would expect, these models are also being used to power AI features across Meta's social platforms. Now, of course, what we didn't get yesterday was any sort of update to Meta's Frontier Level Model 405B, which was released back in August. Meta said while these models referring to the 405B, are incredibly powerful, we recognize that building with them requires significant compute resources and expertise. We've also heard from developers who don't have access to these resources and still want the opportunity to build with Llama. Point being that the competition is not just at the state of the art, but to build a full ecosystem of tools that are applicable for different cost contexts and different needs. When it comes to benchmarking, Meta presented Llama 3.2 as competitive with leading foundation models for image recognition and visual reasoning tasks. Meta's results indicated their models are on par with Anthropic's Claude 3 Haiku and OpenAI's GPT4O mini. They also claim that the smaller text only 3B model outperformed Google's Gemma 2.2.6B and Microsoft's 3.5 mini. Now, of course, part of the place that Meta has staked out in the AI ecosystem is as the great defender of open source. And while the vast majority of developers that I see interacting on Twitter X take Zuckerberg and Meta at their word and with their actions and as truly committed to open source, there are some who are more skeptical. TechCrunch, for example, wrote implicit in Meta's rhetoric is a desire that these tools and models be of Meta's making. Spending on models that it can then commoditize forces the competition to lower prices, spreads Meta's version of AI broadly, and lets Meta incorporate improvements from the open source community. Make no mistake, Meta's playing for keeps. It's spending millions lobbying regulators to come around to its preferred flavor of OpenAI, and it's plowing billions into servers, data centers and network infrastructure to train future models. To be honest though, all that really says is that there is a business strategy behind open source as well. And of course there is. Meta is a huge company. Zuckerberg's an incredibly savvy entrepreneur. The fact that there are benefits to open source, such as the mass proliferation of Meta related models, doesn't to me a priori suggest that the intention behind open source is just a cold calculating capitalism. It was also clear that Meta isn't just thinking about tech specs, they're also thinking about how they integrate these models into products. In their presentation, they showcased a number of new use cases enabled by Llama 3.2 across their platforms, which ironically or perhaps not ironically comes just days after OpenAI fully rolled out. Its advanced voice mode is that Meta writes, you can now use your voice to Talk to Meta AI on Messenger, Facebook, WhatsApp and Instagram DM and it'll respond back to you out loud. Basically, Meta confirms that this sort of voice based interaction where the AI talks back to you is going to be a part of the future, at least assuming consumers actually respond well to it. One specific strategic decision they made was to include celebrity AI voices as some of the options, including Awkwafina Dame, Judi Dench, John Cena, Keegan, Michael Key and Kristen Bell. But a why not? And b if nothing else, it shows the continued intersection of the entertainment industry and the AI space. Meta also highlighted the power of AI to interact with image processing. They gave examples like being able to identify a flower a user sees during a hike or returning a recipe based on an image of a cake. Like all of the big consumer facing AI, they're also showing basic image editing use cases like removing backgrounds. They're also starting to test integrated translation. I speak about this frequently as one of the most obvious but still incredibly cool uses of Gen AI. They write with automatic dubbing and lip syncing. Meta AI will simulate a speaker's voice in another language and sync their lips to match. They are initially rolling out small tests on Instagram and Facebook, translating between English and Spanish with an intent to go broader as soon as they can. While a lot of the focus was on the consumer side, they didn't ignore businesses either. One of their main business use cases is click to message ads on WhatsApp and Messenger, allowing small businesses to set up business AIs that can talk to customers. These tools are seeing uptake. Meta says that more than a million advertisers use the tools and created more than 15 million ads with them in just the last month. They also said that ad campaigns that use Meta's Gen AI features resulted in an 11% higher click through rate and 7.6% higher conversion rate, which is incredibly meaningful at mass scale. Today's episode is brought to you by Plum Generative AI promises to supercharge your productivity and give you superpowers. But if you're not an engineer, trying to harness AI can be incredibly frustrating. Hours wasted wrestling with complex tools only to give up when they don't work. We all have tedious tasks we'd love to automate and challenges AI could solve, but few of us have the skills to fully leverage these game changing technologies. That's where Plum comes in. The mission to make automating your work feel like magic. Imagine typing out AI Read my Gmail and ping me in Slack when something critical comes in and watching it come to life before your eyes. No coding required. Whether you're a marketer, salesperson or founder, Plum enables you to create custom AI workflows in minutes, not hours. Check out useplum.com that's Plum with a B for early access to the future of workflow automation. Today's episode is brought to you by Venice. The leading AI companies store your entire conversation history and attach it to your identity. Free forever. That's every question you ask, every answer you receive, every image you generate, every thought you share with the machine. It's all being spied on. If you trust all the company's hackers and NSA board members that will ever have access to your AI conversations, then rejoice for you are well served. For the rest of us, Venice is an alternative. Venice is a powerful AI app for text, image and code generation that respects you as a sovereign individual and believes privacy and free speech are not only human rights but necessary for civilizational advancement. Private, permissionless and uncensored, you can try it for free without an account. AI Daily Brief listeners receive a 20% discount on Venice Pro. Visit Venice AI NLW and enter the discount code NLW Daily Brief. That's NLW Daily Brief. All one word. Today's episode is brought to you by Superintelligent, which is of course our platform that helps you learn how to use AI tools and practice. Perhaps even more importantly, gives you ideas on the best use cases that are actually going to help you achieve whatever it is you want to achieve. To recognize the end of summer and back to school, back to work. We are running our best promotion ever. When you sign up for Super Intelligent using code Sewback, your first month will be 100% free. The platform features over 600 fun, highly practical AI tutorials that get you using AI fast and with an eye to actually transforming how you get things done. We've just launched super for Teams. So if you have a group of people at your company that want to figure out how to use AI together, I highly suggest you check it out. But for those of you who are using Super Intelligent as an individual once again, if you sign up for Super Intelligent between now and the end of the month using code so back, you will get your first month 100% free. Go to be super AI and check it out today. Now when it comes to community reactions, people are still just wrapping their heads around it. As I mentioned before, there's a lot of positivity around the open source side of things. Dr. Jim Fan from Nvidia writes, I just pulled the numbers on Vision language benchmarks for LLAMA 3.2 11b. Surprisingly, the open source community at large isn't behind in the lightweight model class. Pixtral, Quentu VL, Momo and intern VL2 all stand strong. OSS AI models have never been stronger. Never bet against oss, never underestimate the combined firepower of so many talents distributed all over the world. Now, speaking all over the world, one place this won't be distributed is the eu. This is not a surprise. It's something that Meta had announced, but as Jonas writes, it's still disappointing to see the EU excluded from accessing a promising open source model. The big announcement wasn't about AI, it was about AR augmented reality and specifically what Meta is calling its first true augmented reality glasses, which it calls Orion. Sorry, OpenAI. If you choose that for GPT5's name, it seems like there might be some confusion. Meta presented Orion as the next great leap in human oriented computing. Similar to the Apple Vision Pro, Orion uses a pair of holographic displays to place 2D and 3D content and experiences into your physical surroundings. The glasses feature eye, hand and neural tracking. They also have a very cool new feature where a wrist monitor can understand hand motions so that you can instruct your glasses what to do without having to talk to yourself in public or be waving your hands around like a crazy person. Now, by way of differentiation, especially from other products that have come out in this similar vein, the big one is that many people feel like these are the first AR glasses that don't totally look like dog poop. Yes, they are a little bit clunky, but ultimately they are still in the realm of normal glasses. That's quite different than some other approaches we've seen, but does reflect something that Facebook has seen a lot of success with, which is their Meta Ray Ban glasses, which have been extremely popular. Zuckerberg was very deliberate in calling these glasses not a headset, no wires, and lightweight enough to wear all day. To the extent that there was a catch, it's that these are actually not yet available for sale. It's still a prototype which will be available for Meta employees and select external audiences, although Meta was clear that quote this is not a research prototype, it's one of the most polished product prototypes we've ever developed and it is truly representative of something that could ship to consumers. Rather than rushing to put it on shelves, we decided to focus on internal development first. We which means that we can keep building quickly and continue to push the boundaries of the technology, helping us arrive at an even better consumer product faster. So it seems like there should not be years between when we've seen this and when we can get our hands on them. Still, there were a few people who did have a chance to test these out. The Verge wrote a very, very long and comprehensive review, basically saying that while yes, this product was still being demoed on guardrails and wasn't fully baked from a consumer perspective, that it was very impressive. And frankly, given how much critique we've seen recently of people releasing products that were half baked just to get them out, I think that although people will be disappointed that they can't get their hands on these right now, they also might respond to the better product ultimately that becomes available. Because of course, people haven't had a chance to get their hands on these things. A lot of the conversation was more speculative in general, and one of the biggest things that I saw over and over again was summed up by Billawal Sidhu, who writes Meta is onto a winning formula, pairing neural wristbands with these next gen AR glasses. It's funny, in an alternative universe this is what you'd think Apple would have shipped versus the Vision pro Nikita Beer. Put it even more simply, Zuck is the new Jobs. Of course, that remains to be seen, but I think for many he got a lot closer yesterday. For now, that is going to do it. For today's AI Daily Brief. Until next time, peace.