Meta engineers seem to have corrected the problem what made CEO Mark Zuckerberg so angry that he spent nearly $15 billion to ‘import’ top exec
Meta engineers have reportedly corrected the issue that reportedly plagued company’s AI effort. According to a report by news agency Reuters, quoting the company’s chief technology officer (CTO) Andrew Bosworth, Meta Platforms’ new artificial intelligence lab has delivered its first high-profile AI models internally this month. At a press briefing on the sidelines of the World Economic Forum’s annual meeting in Davos, CTO Bosworth said that the models built by Meta Superintelligence Labs team, formed last year, showed a lot of promise.“They’re basically six months into the work, not quite even,” Bosworth said, adding that the team’s AI models were “very good”. Bosworth did not specify which models were delivered internally. Meta CTO said the technology was not yet finished. Speaking in general terms about the development cycle, Bosworth said: “There’s a tremendous amount of work to do post-training” for AI, “to actually deliver the model in a way that’s usable internally and by consumers.” He said, however, that Meta was starting to see favorable returns from its big gambits in 2025, which he called a “tremendously chaotic year” for Meta building its lab, infrastructure, and procuring power.Bosworth said 2026 and 2027 would see consumer AI trends firm up because recent advances had delivered models that already answer “the kinds of things that you ask every day with your family, your kids,” even as technological development could improve results for more complex queries. Media reports in December said that Meta is working on a new AI model, codenamed Avocado, that could be proprietary instead of open source. Reports claim that the text AI model codenamed Avocado is slated for a first-quarter 2026 launch, along with an image and video-focused model codenamed Mango.
Meta’s AI fail that made Mark Zuckerberg furious
Early last year, Meta CEO Mark Zuckerberg reportedly became intensely frustrated at Meta’s failure to keep pace in the “AI race” with the likes of OpenAI’s ChatGPT and Anthropic’s Claude. Though once set to rival the other America-developed large language models (LLMs), Meta’s LlaMA hit the rocks in May, after its latest “lightweight” models received a disappointing reception. Zuckerberg was reportedly already “furious” that his rivals were enjoying more success. Meta faced criticism over the performance of its Llama 4 model at a time when rivals, such as Alphabet’s Google and others, have seized momentum in the race for transformational and lucrative AI. In early 2025, Meta’s Generative AI unit launched Llama 4, a family of AI models that failed to match the capabilities of Chinese AI lab DeepSeek’s models and that was broadly seen as a disappointment. CEO Mark Zuckerberg went on to shake up Meta’s AI leadership team, form a new lab and poach talent with sky-high offers. Meta spent nearly $15 billion in the data-labeling company Scale AI, taking a 49% stake in the startup, while also bringing on its 28-year-old CEO Alexandr Wang to help lead a new “superintelligence” lab within the company. Zuckerberg had made AI his company’s top priority for the year 2025.
