Mark Zuckerberg ‘doesn’t believe’ AI causes job loss

“People think AI job theft is inevitable, but I don’t really believe this is happening,” Zuckerberg said in the podcast Idea Generation belong to Complex earlier this week.

CEO Meta did not explain his comments in depth, but said that the employment situation would improve if companies focused more on “personal super intelligence” instead of “automating all work with artificial intelligence” (automating all knowledge work). He assessed some businesses trying to create tools to automate work using artificial intelligence as “not good”.

“If there’s a balance between a group of AI companies focused on helping businesses operate more efficiently and another group pursuing the vision of personal superintelligence, which is giving each person more capabilities and gradually helping them work better, I think that would be a pretty positive scenario,” Zuckerberg said.

According to the Facebook boss, businesses always try to operate more efficiently. Therefore, as AI develops, workers also need to keep up with the pace of automation, and this will only happen if they are empowered and tooled.

“If you focus more on empowering people, helping them get better at their jobs at a faster rate, in theory, the future will have more jobs instead of less,” Zuckerberg explained.

Theo Business InsiderZuckerberg appears to be trying to promote Meta’s vision of developing a super personal assistant. The company has spent tens of billions of dollars to attract AI talent and change its artificial intelligence strategy recently, even calling this a “restart of everything”.

Previously, some technology leaders also said that AI does not “steal” jobs, but workers must be equipped with skills related to this technology. In mid-June, Nvidia CEO Jensen Huang said society needs to change to accompany the development of AI, calling on everyone to use this technology. OpenAI CEO Sam Altman also sees AI as just an excuse for businesses to fire employees.

Meta’s AI lags behind despite spending hundreds of billions of dollars

In early April, Meta Superintelligence Labs (MSL) led by billionaire Alexandr Wang launched the large language model (LLM) Muse Spark including a “contemplation mode” that helps coordinate multiple agents at the same time for many different purposes, marking the first time this department released a new AI after more than a year of establishment. Zuckerberg said at the time that Muse Spark was the “first milestone” towards providing personal super intelligence to everyone.

“We’re building products that not only answer your questions, but also act as assistants that do everything for you,” Zuckerberg wrote on Threads at the time. “I’m optimistic that this will support a wave of creativity, entrepreneurship, growth and health.”

Despite what is advertised, Meta and Zuckerberg’s AI ambitions are considered far behind their competitors. Chamath Palihapitiya, a former Facebook director, said Meta is “unlikely” to catch up with rivals in the generative AI race.

“I don’t understand nor understand the motivation behind the question: Why did they fail so miserably,” Palihapitiya told Axios. “But currently, the ambition has not yet been fulfilled.”

Theo FTthe inability to fully deploy Meta’s AI capabilities concerns Google, as the partner is said to be unable to meet its huge computational resource demands, including limiting access to Gemini. CNBC Sources said that as early as March, Meta did not receive the full power Gemini had purchased. The gap between huge demand and limited supply has slowed down Zuckerberg’s company’s internal AI development efforts.

According to the source, other Google customers were also affected, but Meta suffered the heaviest damage due to huge demand. Faced with this situation, Facebook’s parent company forces employees to use tokens “with caution”. Token is the basic unit of AI, consisting of text, code or data that an artificial intelligence model processes or generates.

Gemini is currently an important tool for Meta in moderating content on platforms such as Facebook, Instagram, Threads… including processing and flagging harmful posts and eliminating scams. According to BloombergGoogle’s AI works more accurately than Llama, while Muse Spark has only been deployed for a short time.

Meta and Google have not yet commented.

Meta employee frustration

Despite being one of the tech giants with the most drastic layoffs recently due to AI, the number of Meta employees has not declined. As of April, the company said the total number of employees had increased 1% over the same period last year, to 77,986 people, and also raised its investment expenditure forecast (CapEx) for the whole year to 125-145 billion USD due to accelerating the construction of AI infrastructure. By May, Meta laid off about 10% of its staff (equivalent to about 8,000 positions) and restructured another 7,000 positions.

However, continuous staff cuts and spending billions of dollars on artificial intelligence are making those who stay insecure. Since April, Meta employees have had to install the MCI (Model Capability Initiative) tool on their computers to record mouse and keyboard movements, part of the company’s larger campaign to build AI agents that work autonomously. MCI works on work-related apps and websites, as well as periodically taking screenshots of on-screen content.

But a month later, Meta employees distributed leaflets in many offices to express their frustration, encouraging people to sign an online petition protesting the company’s installation of tracking software. More than 1,600 employees are said to have signed on, and Meta later relaxed the program, allowing employees to pause data collection for up to 30 minutes or apply for exemptions in certain cases.

On June 22, Meta temporarily suspended MCI. According to Reutersthe decision was made after an employee submitted a high-priority security incident (SEV) report regarding data exposure. Exposed data includes commands, logs, private conversations, people and performance data. The incident is rated SEV 2 on a scale of 0 to 5, with 0 being the most severe.

“While there is no indication that Meta employees illegally accessed data, we will temporarily pause this activity to conduct an investigation,” a Meta spokesperson told AFP June 30.

Theo AFPMeta staff is still in a state of “anxiety and anger”. One person even summed up the mood with a meme from The Office – a movie about workplace satire – posted on an internal forum with the content: “0 days since our last silly act”.