Zuckerberg Says Meta Will Add Generative AI To All Its Products

Meta is in furious competition in the rapidly expanding generative AI space with the likes of Google, Microsoft and OpenAI, and that includes trying to find talent in a relatively shallow pool. Given that, the parent company of Facebook and other products reportedly is setting aside prior practices to keep employees it has and lure others from its rivals.

According to a report in The Information, that includes CEO Mark Zuckerberg reaching out to researchers at Google’s DeepMind AI unit via personal email messages encouraging them to come to Meta. Sources showed the news site copies of some of those emails.

The company also has taken other steps, from offering jobs to candidates without requiring an interview and reversing its policy of not offering employees being offered jobs at other organizations salary increases in hopes of convincing them to stay, according to the report.

The push to ramp up recruitment comes amid reports of AI researchers leaving Meta for other companies that reportedly tend to pay more, with some jobs at organizations like OpenAI offering $5 million to $10 million in stock options for top AI talent.

Meta’s Expansive AI Plans

It’s also a time when Meta, like other high-profile companies, is looking to build out its AI capabilities and integrate them throughout its product and services portfolio. Zuckerberg in January on Instagram talked about plans to build out its AI infrastructure to support artificial general intelligence (AGI) – the theoretical stage at which AI can think and reason as well as humans – which included spending billions of dollars to buy 350,000 of Nvidia’s H100 GPUs, the go-to chip for AI training and inferencing workloads. Adding in other GPUs, the plan is by the end of the year to have the equivalent compute capacity of 600,000 H100s.

It also is working on the latest generation of its in-house custom chip, “Artemis,” with plans to deploy it in its data centers this year. The new chips would be used for inferencing workloads and, the company hopes, reduce its costs. In addition, Meta will use Nvidia’s new Blackwell GPUs, announced earlier this month, but due for release later this year, to build and train its Llama large language model (LLM).

Meta has already integrated some of its AI into its Facebook, Messager, Instagram and WhatApp platforms, though recent reports indicate the company is working on a new feature for WhatsApp that will let users ask Meta AI questions from the app’s search bar.

In addition, earlier this month, Meta executives said they were creating an AI system to power Facebook’s entire video recommendation engine.

All of that will take an infusion of talented AI employees. Meta, which laid off tens of thousands of employees in 2022 and early 2023, said in October that it planned to grow employment this year to “support priority areas … which we expect will continue to shift our workforce composition toward higher-cost technical roles.”

Finding AI Talent Isn’t Easy

Given the tight market for AI talent, that will be a challenge. Myriad studies and surveys show that the skills gap in AI continues to widen even as innovation around the technology accelerates. Software vendor SnapLogic in a study found that 93% of respondents in the United States and the UK said AI was a business priority, but 51% said they don’t have the in-house skills to execute on their plans.

In addition, the lack of talent was cited by 40% of respondents as the number-one barrier to moving forward with their strategies.

A report by Salesforce last year found that 62% of workers said they didn’t have the necessary skills to effectively and safely use generative AI and 70% of executives didn’t believe their employees had those skills.