CALDEIRÃO DE BOLSA

MarcoAntonio Escreveu:Isto tem sido bastante mais discutido nos últimos meses, não só por uma série de processos que foram colocados entretanto, mas também porque diversos leaks e whistleblowers têm indicado que estas techs estão completamente conscientes do comportamento adictivo (apesar de publicamente o negarem).

‘We’re basically pushers’: Court filing alleges staff at social media giants compared their platforms to drugs

Google, Meta, push back on addiction claims in landmark social media trial

Investigação interna por parte das próprias tech:

Facebook negatively impacts sleep, work, relationships for about 1 in 8 users, documents show

Meta’s own research found parental supervision doesn’t really help curb teens’ compulsive social media use


Isto já vem de 2021:

The Facebook Whistleblower Revealed Herself on 60 Minutes. Here’s What You Need to Know

She submitted tens of thousands of those documents to the Securities and Exchange Commission and to Congress. She was also the primary source of the Journal’s Facebook Files stories last month that exposed how Facebook’s algorithm fosters anger to drive engagement, how it favors high-profile users, and how Instagram made nearly one-third of teenage girls who use it feel worse about their bodies.

Para lá até da questão do comportamento adictivo, são revelações como estas que têm estado por detrás das cada vez mais frequentes proibições das redes sociais a crianças e adolescentes.

She submitted tens of thousands of those documents to the Securities and Exchange Commission and to Congress. She was also the primary source of the Journal’s Facebook Files stories last month that exposed how Facebook’s algorithm fosters anger to drive engagement, how it favors high-profile users, and how Instagram made nearly one-third of teenage girls who use it feel worse about their bodies.

As ações da dona do Facebook caíram 16% nos últimos 12 meses devido aos receios de que a empresa esteja a gastar demasiado em inteligência artificial. O fundo considera que as preocupações estão a subestimar o potencial de crescimento a longo prazo.

Mark Zuckerberg quer aumentar a sua posição no mundo de inteligência artificial e traçou grandes contratos para consumir e pagar energia nuclear, potencializando os seus negócios.

Meta Superintelligence Labs is already losing key staff less than two months after launch

• Some of Meta's AI staff are exiting its new superintelligence division.
• These departures follow Meta's push to compete with OpenAI and Google.
• Meta says the departures are mostly from longtime AI employees, and some attrition is "normal."

Meta's high-stakes bet on "superintelligence" is showing early cracks.

At least eight employees, including researchers, engineers, and a senior product leader, have left the company less than two months after CEO Mark Zuckerberg announced a brand new division called Meta Superintelligence Labs (MSL) aimed at bringing "personal superintelligence" to everyone.

The departing employees are mostly veterans who helped build some of Meta's core AI infrastructure, along with some newer hires lured by Zuckerberg's pitch to join MSL.

(...)

Wu told Business Insider that some Meta AI workers felt work was unstable at times due to constant reorganizations.

"Speaking generally and not for myself, a lot of people in the AI team maybe feel things are too dynamic," he said. "There were a lot of organizational changes — actually, my manager changed several times."

(...)

Some longtime Meta veterans are leaving

Bert Maher, who spent 12 years at Meta and was deeply involved in building some of its most important AI tools, left earlier this week to join Anthropic. At Meta, Maher helped to develop PyTorch, an open-source software that has become one of the most widely used tools for training and testing AI systems. He also worked on Triton, a programming language and compiler for making AI models run more efficiently. In a post on X, Maher said that he was excited to make Claude, Anthropic's AI chatbot, "even faster." Maher declined to comment.

(...)

Some recent Meta hires left, too

The recent departures mostly come from the ranks of employees who joined Meta well before it announced its high-profile superintelligence push, though some also include recent hires.

Wired reported that two researchers, Avi Verma and Ethan Knight, left MSL after less than a month and returned to OpenAI, where they had previously worked. Verma left MSL before his start date, a person familiar with the matter told Business Insider. Verma and Knight did not respond to requests for comment.

"During an intense recruiting process, some people will decide to stay in their current job rather than starting a new one. That's normal," the Meta spokesperson said of the new hires.

Rishabh Agarwal, who joined Meta from Google DeepMind in April, decided to leave after just five months. In a post on X, Agarwal praised Meta's "talent and compute density" and said that he wanted to take "a different kind of risk."

According to a person familiar with the matter, Agarwal is joining Periodic Labs, a new AI startup founded by former OpenAI and DeepMind researchers that is using AI to study and discover new physical materials.

Meta to license AI technology from start-up as in-house models lag rivals

Partnership with Midjourney marks shift away from internal product development



Meta will license technology from artificial intelligence image and video generation start-up Midjourney, as the social media group shifts towards working with third parties as its struggle to keep pace with rivals.

(...)

The tie-up will allow Meta to develop and integrate multimedia AI generation features into its apps, as chief executive Mark Zuckerberg has indicated that he expects AI-generated content to become more prominent on the platform.

(...)

The Midjourney partnership marks a shift by Meta away from building all of its AI models and products in house, after its existing ones began to lag rivals.

In 2024 Meta rolled out an image generation tool called Imagine, which allows users to generate images from text prompts. Last October it shared a research paper on a movie generation model, Movie Gen, that will generate and edit videos based on text prompts. Meta promised to integrate it fully into Instagram in 2025.

However, the integration has yet to happen and industry insiders say the model already appears antiquated compared with Google’s Veo 3 and OpenAI’s Sora models, which have been released to consumers.

The social media company had also abandoned plans to publicly release its flagship Behemoth large language model, according to people familiar with the matter, focusing instead on building new models.

Meta had started using third-party models internally for tasks such as coding, according to multiple people familiar with the matter, as faith in its Llama models has waned.

(...)

Meta Freezes AI Hiring After Blockbuster Spending Spree

Company reorganizes its sprawling artificial-intelligence operation, whose spiraling cost has drawn investor scrutiny


Meta Platforms has frozen hiring in its artificial-intelligence division after spending months scooping up 50-plus AI researchers and engineers, according to people familiar with the matter.

The hiring freeze, which went into effect last week and coincides with a broader restructuring of the group, also prohibits current employees from moving across teams inside the division. The duration of the freeze wasn’t communicated internally.

There might be exceptions to the block on external hires, but they would need permission from Meta’s chief AI officer, Alexandr Wang, the people said.

A Meta spokesperson confirmed the freeze, characterizing it as “basic organizational planning: creating a solid structure for our new superintelligence efforts after bringing people on board and undertaking yearly budgeting and planning exercises.”

While all of the top AI companies have hired aggressively this year, Meta has most often pushed the pace of the talent war, offering prized researchers pay packages worth nine figures and using so-called reverse acquihires to strip startups of key leaders. Analysts have voiced concerns about the scale of leading tech firms’ investments, with some singling out Meta’s fast-rising stock-based compensation costs as a potential threat to shareholder returns.

The recent restructuring inside Meta divides its AI efforts into four teams: one working on superintelligence, called TBD Lab, that houses many of the new hires; a second working on AI products; a third working on infrastructure; and a fourth dedicated to projects with a longer time horizon and more exploration, the people said. The latter, called Fundamental AI Research, remains largely untouched in the reorganization.

The four groups sit within the umbrella of Meta Superintelligence Labs, a name that reflects Chief Executive Mark Zuckerberg’s recent emphasis on building computer systems that can outperform the smartest humans on cognitive tasks. The Information previously reported some details of Meta’s AI reorganization.

Previously, Meta had a team called AGI Foundations, which worked on the latest versions of the company’s large language models, known as Llama. The team drew criticism from executives this spring after the release of the latest Llama models underperformed expectations. It was dissolved as part of the reorganization.

At least three members of the previous AGI Foundations team said internally that they were leaving the company around the time of Meta’s latest vesting date, Aug. 15, according to internal posts viewed by The Wall Street Journal.

It was after the Llama model release in April that Zuckerberg became personally involved in recruiting AI researchers. He approached employees of OpenAI, Google DeepMind and other labs with email overtures and WhatsApp messages that quickly led to offers that in some cases reached $100 million in total compensation. His offer to one researcher—Thinking Machines Lab co-founder Andrew Tulloch—could have been worth as much as $1.5 billion. (Tulloch declined the offer.)

To spearhead Meta’s AI efforts, Zuckerberg hired Scale AI co-founder Wang, securing him by paying $14 billion for a stake in his company. He also brought on former GitHub CEO Nat Friedman and Safe Superintelligence co-founder Daniel Gross, whose courtship involved Meta offering to buy a stake in their venture firm.

As of mid-August, Meta had successfully hired more than 20 researchers and engineers from OpenAI for the effort, at least 13 from Google, three from Apple, three from xAI and two from Anthropic for a total of 50-plus new employees.

Mounting concern from investors over the costs of the tech giants’ AI buildout has played a role in this week’s selloff of technology stocks. In an Aug. 18 research note, analysts at Morgan Stanley warned that the fast-rising stock-based compensation offered by Meta and Google to lure AI talent could threaten their ability to return capital to shareholders via buybacks. Lavish spending on talent, the analysts wrote, “has the potential to drive AI breakthroughs with massive value creation or could dilute shareholder value without any clear innovation gains.”

A dona do Facebook está a começar uma reestruturação na área da inteligência artificial. Com a divisão da unidade em quatro frentes de ataque, Mark Zuckerberg pausou a procura por novos talentos.

Com mercados a temerem bolha na IA, Zuckerberg volta a reorganizar o núcleo do projeto da Meta na área.

Mark Zuckerberg não tem poupado esforços para colocar a Meta na linha da frente da revolução da inteligência artificial. Nos últimos meses, o criador do Facebook atraiu investigadores com salários de nove dígitos, investiu mais de 14 mil milhões de dólares na Scale AI e nomeou Alexandr Wang, fundador desta start-up, como novo “chief A.I. officer”. Mas agora, depois de meses de contratações em massa, chega a fase da “limpeza”.

A informação foi avançada pelo The New York Times esta terça-feira. Segundo o jornal norte-americano, a Meta vai dividir a sua unidade, a Meta Superintelligence Labs, em quatro frentes distintas: investigação fundamental; superinteligência; produtos e infraestruturas. Ao mesmo tempo, a empresa prepara-se para reduzir o peso da equipa, que já conta com vários milhares de pessoas. Isso poderá significar cortes ou a transferência de colaboradores para outras áreas. Até agora, não há decisões finais sobre o futuro dos colaboradores, embora o The New York Times confirme que o debate está a decorrer com “fluidez”.

A mudança acontece após fricções internas resultantes da chegada de novos executivos. Alguns investigadores de longa data mostraram desconforto com a pressão imposta pela nova geração de líderes. Shengjia Zhao, antigo cientista do OpenAI e co-criador do ChatGPT, foi recentemente nomeado “chief A.I. scientist” da Meta e tem reavaliado o trabalho das equipas históricas. “Tem havido uma fila de investigadores veteranos à porta do seu gabinete, a serem entrevistados para novas funções”, descreveu uma fonte ao jornal.

A Meta também está a repensar a sua filosofia. Depois de anos a apostar em modelos “open source”, como o Llama, há discussões internas sobre a adoção de sistemas “fechados”, guardando em segredo o código dos modelos mais avançados. A decisão significaria um afastamento claro da cultura de abertura que a empresa promove.

A reorganização surge ainda após o fracasso do “Behemoth”, um modelo de IA que falhou nos testes de desempenho e acabou por ser abandonado. Agora, a equipa liderada por Alexandr Wang, procura criar de raiz um novo “frontier model” que represente a grande aposta de Zuckerberg na chamada superinteligência que, segundo o próprio, poderá inaugurar “uma nova era de capacitação individual”.

(...)

A Constellation Energy concordou em vender energia de uma central nuclear em Illinois à dona do Instagram. O contrato a 20 anos espelha o apetite das gigantes de tecnologia em operar centros de dados em inteligência artificial.

BearManBull Escreveu:

Facebook and Instagram get rid of fact checkers

Um pouco de ar fresco.

Facebook and Instagram get rid of fact checkers