Welcome to Lowpass! This week: What three months of Netflix job listings tell us about the company’s plans for LLMs and generative AI.

Netflix is staffing up to use AI everywhere

“[There is] lots of hype, good and bad, about how AI is going to impact or transform the entertainment industry. […] And it's very important, I think, for creators to be very curious about what these new tools are and what they could do.”

When Netflix co-CEO Ted Sarandos was asked about the impact AI will have on his company’s business during the company’s Q3 earnings call earlier this month, he seemed optimistic, with a few caveats. “AI needs to pass a very important test,” Sarandos argued. “Can it help make better shows and better films?”

That’s a question Hollywood has been grappling with for some time now. Creatives are spooked by the idea that studios could use AI to replace background actors, draft mediocre scripts and pay human writers less to rework them, and find other ways to cut corners. That’s why AI has been at the center of recent Hollywood labor strikes, and that’s perhaps also why Sarandos treaded carefully on the subject, suggesting that his company wanted to use new technologies to make shows and movies better, not cheaper.

“Any tool that can go to enhance the quality, making them better is something that is going to actually help the industry a great deal,” Sarandos concluded.

For Netflix, this is not just a mere thought exercise. I have been analyzing three months of Netflix job listings, and found that the company is exploring the use of LLMs and generative AI models across its entire organization for a variety of tasks ranging from live captioning and personalization to content production. Netflix’s plans include building an internal AI platform that can be used across the company to build generative AI applications.

Netflix has long used machine learning, and now embraces LLMs. Netflix was a tech company long before it became a streaming and Hollywood powerhouse, and has used AI and machine learning in a variety of contexts for a long time. Any time you open the Netflix app and browse its recommendations, you’re exposed to a personalized experience powered by machine learning.

However, the company now appears to fully embrace the latest advancements in artificial intelligence, including LLMs and generative AI models. This includes using LLMs for search and personalization, according to multiple recent job listings.

• “Fast-paced innovation in Generative AI and large language models (LLMs) is now advancing numerous fronts of the Search and Recommendations experiences on the Netflix product, including content discovery and personalization for members,” details one job listing, which also promises potential applications that they will be “at the forefront of research and application of LLM innovation, NLP, and Machine Learning.”

• “We are looking for an exceptional applied research engineer to help us develop the technology to power future search and recommendations experiences using the latest advances related to LLMs,” notes another job listing.

Netflix is exploring the use of AI for captioning and localization. As a global streaming service with worldwide hits like Squid Game and Casa de Papel, Netflix has long heavily relied on captioning, dubbing and UI localization to speak to its audience in many different languages. Now, it is looking to AI to supercharge those efforts.

• “We are looking for an experienced Research Scientist with established track records in Multimodal LLM research areas to develop algorithms that power high quality localization at scale,” one recent job listing states. “In this rare opportunity, you will have the chance to conduct cutting-edge Machine Learning research to help enhance localization and global entertainment at Netflix.”

• One particular focus appears to be using LLMs for live captioning, with one recent job description specifically mentioning that applications will have to “research, develop and productize cutting-edge Multilingual Live Caption and Translation models.”

Netflix also wants to use AI in content production. As I said, this is a touchy subject, and it’s worth noting that nothing in Netflix’s recent job listing suggests that the company is getting ready to replace actors or extras with AI. However, Netflix is also not shying away from exploring the use of AI in production altogether.

• This includes researching ways creatives can use AI within their own workflows. Netflix has for years built a number of apps to change how Hollywood works, which includes moving people away from paper schedules. One recent job listing suggests that it wants to explore ways to tap AI for such tools in order to “unlock new creative possibilities for storytellers and drive efficiency across the content creation process.”

• If that sounds a bit vague, then here’s something more concrete: One recent job listing specifically mentions the use of generative AI for CGI, and tells applications that they will “design and develop tools that support artists' workflows by integrating GenAI models to expand creative opportunities.” Applicants are being told that their chances will improve if they have “experience with models and tools for image/video generative AI.”

Netflix is building its own AI infrastructure. To power all these different use cases, Netflix isn’t just going to fire up GPT and call it a day. Instead, the company has been building its own AI platform, according to multiple job listings.

• One job listing explains the role of this platform this way: “We take care of the heavy lifting needed to build compliant and innovative generative AI-powered experiences that support the expanding needs of the business,” adding that a still-to-be-hired product manager for the platform will “lead the vision and strategy for our model consumption layer, conversational assistants, LLMOps, prompt management, prototyping tools, and an opinionated app development framework for AI applications and agents.”

• Another listing details that an engineer working on the platform will “design, build, and deploy APIs to access third-party large language models and internal fine-tuned ML models” and also “develop frameworks and components for building GenAI applications.”

• At least some of these applications will not be consumer-facing. One role Netflix is looking to fill is on the N-Tech team, which builds internal enterprise apps. Applications will “apply AI and GenAI responsibly to solve enterprise problems for the whole company,” the listing states.

Everyone is in love with AI these days, and it’s no surprise that a company the size of Netflix is exploring ways to use LLMs and generative AI. However, you have to look at any move the company is making in this space in the context of Netflix’s tech DNA. The company was using machine learning for curation before most of its competitors even had a streaming business, and its engineering workforce dwarfs that of anyone else in Hollywood to this day. For Netflix to now aggressively hire in this field means that it could once again outrun the competition.

How I stay up to date on cord cutting trends

If you’ve followed my work for a while, you may know that I used to write a lot about the practicalities of cord cutting. A decade ago, I even hosted a YouTube show called Cord Cutters … fun times!

One interesting aspect about this more consumer-focused coverage — device reviews, cord cutting tips and so forth — was that I got a surprising amount of feedback from industry insiders. This makes sense: At the end of the day, studio execs and streaming service engineers watch TV as well, after all. But even more so, seeing what services are out there, how devices work and how consumer actually watch TV helps to understand where the industry is going, and what opportunities others might be missing.

These days, I don’t write a whole lot of this kind of how-to content anymore, but I still like to stay up to date on the latest and greatest in consumer cord cutting. For that, I’ve long valued Jared Newman’s free Cord Cutter Weekly newsletter. Jared not only writes about the latest streaming services and devices, but also features some great ways to save money in every issue. Go check it out!

What else

Snap’s new Spectacles are coming to Europe. AR developers in Germany, France, Spain, Italy, Austria, and the Netherlands are now able to rent the glasses from Snap for €110/month. Here’s my recent story about the new Spectacles.

Nintendo now has a music streaming service. Surprise: Nintendo Music, which is now available on Android and iOS, only plays Nintendo Music. 

Meta has invested more than $58 billion into AR/VR to date. The company’s Reality Labs unit generated $270 million in revenue in Q3, up 29% year-over-year.

Amazon’s Echo turns ten. The Verge is celebrating the anniversary with a package of stories worth checking out.

However, the new Alexa is missing the birthday party. The launch of an LLM-powered version of the voice assistant has been delayed until 2025, reports Bloomberg.

Google gears up for an Android XR launch. Android Authority found mentions of headsets in the latest Google Play Store app.

Netflix wants people to share its clips. The streaming giant just made it easier to share clips on social media.

Tidal is doing mass layoffs. Cuts of up to 25% of its workforce are supposed to help Tidal operate “like a startup again,” says parent company owner Jack Dorsey. (Insert your obligatory “beatings will continue until morale improves” reference here.)

That’s it

I asked Meta AI for Halloween costume inspirations. My idea: I want to go as a newsletter writer. Meta AI’s response: Prepare for battle.

Thanks for reading, have a great weekend!