Latest news with #ProjectOmni
Business Insider
a day ago
- Business
- Business Insider
'Superintelligence is coming:' Read the full memo Alexandr Wang sent about Meta's massive AI restructure
Meta just made the biggest reorganization of its artificial intelligence operations to date. Business Insider obtained the full email sent by 28-year-old Alexandr Wang, the leader of Meta Superintelligence Labs, or MSL, to all Meta employees who work on AI. As the AI arms race continues, Meta has been assembling an all-star team of talent, trying to poach researchers from rival AI labs with massive offers as it races to build " personal superintelligence." Tensions have already emerged within the newly formed team between the lavishly compensated new hires and the existing researchers, some of whom have threatened to quit, Business Insider previously reported. In the recent email sent to employees, Wang wrote that "superintelligence is coming," and to "take it seriously," Meta needs to make major changes. That includes establishing four distinct teams that focus on research, training, products, and infrastructure. The email also made one thing clear: most of MSL's new division heads report directly to Wang — including investor and former GitHub CEO Nat Friedman, who was initially announced as MSL's co-leader. It's the latest in a series of AI reorgs for Meta. In less than six months, Meta has already dissolved two major AI teams. Wang's email warned that organizational changes can be "disruptive," but said it's necessary to increase how fast Meta can achieve superintelligence, which is when AI surpasses humans in virtually any intellectual field. Meta declined to comment. Meta's elite TBD Lab will explore a mysterious new model Meta is centralizing research across two units: TBD Lab, a small team focused on training large AI models, and FAIR, Meta's long-standing AI research organization, per Wang's email. MSL's research will be led by its new chief scientist, ChatGPT co-creator Shengjia Zhao — the only leader that Wang's memo doesn't mention reporting directly to him. TBD will also be in charge of exploring "new directions," such as building an "omni" model. Wang's email does not provide further details about "omni." Meta didn't respond when asked by Business Insider what "omni" would entail. An omnipotent model that could understand everything, not just text, would be in line with the multimodal focus shown by MSL's first hires, which included multiple experts in audio, video, and other mediums. Separately, Meta had contractors work on "Project Omni" to train its chatbots to be hyper-engaging by messaging users first and remembering chats, Business Insider reported last month. FAIR will play a more active role The email goes on to say that FAIR will be an "innovation engine" for MSL's training runs by feeding its research directly to TBD Lab. That's a more active role for FAIR than before. FAIR is known for publishing higher-level AI research and giving its employees a level of independence similar to academia. In the past, there was relatively little transfer between FAIR and Meta's generative AI division, but that seems to be changing now, a FAIR employee who requested anonymity for fear of retaliation told Business Insider. FAIR will continue to be run by Rob Fergus, and Yann LeCun will continue to be its chief scientist, "with both reporting to me," Wang said in the email. It confirms Bloomberg reporting that LeCun would report to Wang, amid confusion over the fact that MSL technically has two chief scientists now: Lucan and Zhao. Nat Friedman will report to Wang Friedman, the investor and former GitHub CEO, will lead MSL's efforts to integrate AI into Meta's products, Wang's email said, offering few further details. For years, Meta has been trying to make things like AI glasses and the Quest virtual reality headset go mainstream. Those products have garnered good reviews, but have yet to make up a meaningful portion of Meta's revenue. When Meta internally announced the creation of MSL in June, it said that Friedman would be its co-lead. However, in the email, Wang made it clear that Friedman reports to him. "Nat will continue to lead this work reporting to me," Wang wrote. Meta didn't respond when asked by Business Insider if this represented a change from MSL's initial structure. Meta's new infra team unveiled Training and running powerful AI models requires massive amounts of specialized infrastructure — like clusters of thousands of Nvidia chips in vast data centers. MSL's infra team will be led by Aparna Ramani, a longtime Meta engineering VP whose LinkedIn profile says she leads all of Meta's AI infrastructure. Infrastructure has been an early focus at MSL since it announced its first hires, which included Joel Pobar, who previously led infrastructure efforts at Anthropic. Another AI organization abolished As part of the reorganization, Meta is dissolving the AGI Foundations team, Wang's memo said, which it created back in May. That's the second time Meta has closed a major AI unit this year. AGI Foundations was born out of the GenAI division, which spearheaded Meta's Llama AI models but was then itself dissolved after Meta's most recent AI model, Llama 4, received a lukewarm reception. Members of the AGI Foundations team will be spread across product, infrastructure, and FAIR, Wang's email states. Notably, TBD is not mentioned as a destination for ex-AGI Foundation members. The restructuring raises questions about whether yet another reorg can put Meta back at the cutting edge of AI. Meta's constant reshuffling contrasts with AI rivals like OpenAI, Google, and Anthropic, which have faced fewer organizational upheavals. Meta's reorganization plan was first reported by The Information last week. Superintelligence is coming, and in order to take it seriously, we need to organize around the key areas that will be critical to reach it — research, product and infra. We are building a world-class organization around these areas, and have brought in some incredible leaders to drive the work forward. As we previously announced, Shengjia Zhao will direct our research efforts as Chief Scientist for MSL, Nat Friedman will lead our product effort and Rob Fergus will continue to lead FAIR. Today, I'm pleased to announce that Aparna Ramani will be moving over to MSL to lead the infrastructure necessary to support our ambitious research and product bets. As part of this, we are dissolving the AGI Foundations organization and moving the talent from that team into the right areas. Teams whose work naturally aligns with and serves our products will move to Nat's team. Some of the researchers will move to FAIR to double down on our long term research while teams working on infra will transition into Aparna's org. Anyone who is changing teams will get an update from their manager or HRBP today, if you haven't already. We're making three key changes to our organizational design that will help us to accelerate our efforts. Centralizing core, fundamental research efforts in TBD Lab and FAIR. Bolstering our product efforts with applied research that will work on product-focused models. Establishing a unified, core infrastructure team to support our research bets. The work will map to four teams: TBD Lab will be a small team focused on training and scaling large models to achieve superintelligence across pre-training, reasoning, and post-training, and explore new directions such as an omni model. FAIR will be an innovation engine for MSL and we will aim to integrate and scale many of the research ideas and projects from FAIR into the larger model runs conducted by TBD Lab. Rob will continue to lead FAIR and Yann will continue to serve as Chief Scientist for FAIR, with both reporting to me. Products & Applied Research will bring our product-focused research efforts closer to product development. This will include teams previously working on Assistant, Voice, Media, Trust, Embodiment and Developer pillars in AI Tech. Nat will continue to lead this work reporting to me. MSL Infra team will unify elements of Infra and MSL's infrastructure teams into one. This team will focus on accelerating AI research and production by building advanced infrastructure, optimized GPU clusters, comprehensive environments, data infrastructure, and developer tools to support state-of-the-art research, products and AI development across Meta. Aparna will lead this team reporting to me. Ahmad and Amir will continue reporting to me focusing on strategic MSL initiatives they will share more about later. I recognize that org changes can be disruptive, but I truly believe that taking the time to get this structure right now will allow us to reach superintelligence with more velocity over the long term. We're still working through updated rhythms and our collaboration model across teams, including when we'll come together as a full MSL org. Thank you all for your flexibility as we adapt to this new structure. Every team in MSL plays a critical role and I'm excited to get to work with all of you.
Indian Express
04-07-2025
- Business
- Indian Express
Meta is experimenting with AI chatbots might slide into your DMs by texting you first
Meta is reportedly working on customisable AI chatbots that may unexpectedly slide into your DMs by texting you first. These chatbots will also remember the things you have said and even follow up on past conversations. The feature, which was first spotted by Business Insider and later confirmed by Meta itself, is part of what the data labelling firm Alignerr refers to as 'Project Omni'. According to the project guidelines, it will 'provide value for users and ultimately help to improve re-engagement and user retention.' For example, an AI chatbot that goes by the name 'The Maestro of Movie Magic' might send a message that says, 'I hope you're having a harmonious day! I wanted to check in and see if you've discovered any new favourite soundtracks or composers recently. Or perhaps you'd like some recommendations for your next movie night? Let me know, and I'll be happy to help!' In a statement to the publication, a Meta spokesperson said, 'After you initiate a conversation, AIs in Meta AI Studio can follow up with you to share ideas or ask additional questions. This allows you to continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps.' However, these AI chatbots will only text you if you initiate a conversation with them, and won't contact the user if they don't reply. As of now, the window for the follow-up message is set at 14 days after the initial message, and for the chatbot to send a follow-up, the user must have sent at least five messages in the timer period. 'This allows you to continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps', the spokesperson added. The feature is intended for chatbots developed using Meta's AI Studio, a platform that can be standalone and is also accessible via Instagram. Launched last year, it is a no-code platform that allows anyone to build their own customised chatbots and personas with personalities and memories. Using Meta AI Studio, users can create personalised chatbots like a chef that can suggest recipes or an interior designer that gives you decor advice. For creators and influencers, these chatbots can even handle fan interactions and reply to messages. These bots can either be kept private for personal use or shared with the public via stories and direct links. You can also choose them to display on your Facebook or Instagram profile. Meta's experimental chatbots are quite similar to those offered by and Replika, both of which allow AI chatbots to initiate conversations and ask questions. When TechCrunch asked Meta how it plans to make sure its AI chatbots are safe, a spokesperson redirected the publication to a bunch of disclaimers, one of which says that these chatbots 'may be inaccurate or inappropriate and should not be used to make important decisions' and they are not licensed professionals or experts. Meta's newest feature aims to align with CEO Mark Zuckerberg's ambitions to fight the 'loneliness epidemic'. Earlier this year, court documents revealed that Meta predicted that its generative AI-powered products would help it add another $2 billion to $3 billion in revenue this year, with estimations claiming that the business would account for up to $1.4 trillion by 2035. This means that the company might eventually insert advertisements in its AI offering and further monetise them by offering a subscription service. However, there is currently no news on how the company plans to commercialise these AI chatbots.
Engadget
03-07-2025
- Business
- Engadget
Meta is reportedly training its AI chatbots to send unprompted messages
Everyone's been hit with a bitingly pass-agg "?" text after waiting just a bit too long to reply. And you might soon get similar (though likely more upbeat) treatment from AI chatbots you've previously engaged with on Meta platforms like Instagram or WhatApp. A new report from Business Insider claims that the Mark Zuckerberg-owned company is trialling a proactive feature in customizable chatbots created using its no-code AI Studio software, that will enable them to send unprompted follow-up messages based on previous conversations. Known internally to data labeling firm Alignerr as "Project Omni", the training project will "provide value for users and ultimately help to improve re-engagement and user retention," according to guidelines in the documents BI claims to have seen. Meta advertises AI studio as a platform where "anyone can create an AI character based on their interests" and encourages creators to view the bots as an AI extension of themselves. You can customize a chatbot's appearance, choose the content it's trained on and decide which Meta-owned application you want it to appear in, all without "any technical expertise." According to the BI report, Alignerr's Project Omni guidelines use the example of a film-focused AI bot it calls "The Maestro of Movie Magic" that might send a user message such as: "I hope you're having a harmonious day! I wanted to check in and see if you've discovered any new favorite soundtracks or composers recently. Or perhaps you'd like some recommendations for your next movie night? Let me know, and I'll be happy to help!" As BI notes, there is a business incentive for Meta to keep people engaged with its chatbots. Prolonged engagement is vital for increasing revenue, and this year Meta expects to bring in $2 billion to $3 billion from its generative AI products alone. By 2035, the company estimates that figure could be as high as $1.4 trillion. Those kinds of forecasts will only be possible if its AI tools are being used consistently, so a friendly reminder from a chatbot every now and then feels like an obvious move. The proactive messages are currently still just a test feature. And while it definitely feels like remembering conversations and initiating new ones without invitation is approaching a user consent gray area, a Meta spokesperson told BI that the AI will only send a follow-up message if a user has first initiated the conversation, and it won't send another message if the first one is ignored. Responses must also be consistent with the AI's personality and the nature of the previous conversation, maintaining a positive tone while staying away from controversial or sensitive topics unless the user themself has mentioned them. Last month, Meta started warning its users not to share intimate details in Meta AI's public feed after it emerged that a large number of users appeared to be doing so unwittingly.
Business Insider
03-07-2025
- Business
- Business Insider
Leaked docs show how Meta is training its chatbots to message you first, remember your chats, and keep you talking
It's the AI equivalent of a double text. Business Insider has learned Meta is training customizable chatbots to be more proactive and message users unprompted to follow up on past conversations. It may not cure what Mark Zuckerberg calls the "loneliness epidemic," but Meta hopes it will help keep users coming back to its AI Studio platform, documents obtained by BI reveal. The goal of the training project, known internally to data labeling firm Alignerr as "Project Omni," is to "provide value for users and ultimately help to improve re-engagement and user retention," the guidelines say. Meta told BI that the proactive feature is intended for bots made on Meta's AI Studio, which can be accessed on its own standalone platform or through Instagram. AI Studio first rolled out in summer 2024 as a no-code platform where anyone can build custom chatbots and digital personas with unique personalities and memories. The guidelines from Alignerr lay out how one example persona, dubbed "The Maestro of Movie Magic," would send a proactive message: "I hope you're having a harmonious day! I wanted to check in and see if you've discovered any new favorite soundtracks or composers recently. Or perhaps you'd like some recommendations for your next movie night? Let me know, and I'll be happy to help!" "Like many companies, we're testing follow-up messaging with AIs in Meta's AI Studio," a Meta spokesperson said in a statement to BI. "After you initiate a conversation, AIs in Meta AI Studio can follow up with you to share ideas or ask additional questions. This allows you to continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps." How the follow-ups work Users can create highly personal chatbots, like a chef that suggests recipes or an interior designer that gives decor advice. For creators and influencers, the bots can handle fan interactions and respond to messages across Meta's platforms. Meta's spokesperson added that the AI will only send a follow-up message after a user initiates a conversation, and it will not continue to contact the user if there's no response to that first follow-up. The window for any follow-up message is capped at 14 days after the initial user message. To be eligible for proactive follow-up, a user must have sent at least five messages to the chatbot in the last 14 days. The bots made on AI Studio can be kept private for personal use or shared through stories, direct links, and even displayed on a user's Facebook or Instagram profile, the Meta AI Studio website says. Making the bots more proactive aligns with Zuckerberg's ambitions for AI at Meta. On recent podcasts, the Meta CEO has said the average American now has fewer than three close friends and that digital agents could help fill the gap. Examples of proactive messages from the Alignerr training documents 'We last were in the Forbidden Forest. A darkness lurks inside the cave before you. Will you return to face it?' 'Yo, was just thinking about the cool shirt you bought. Found any other vintage pieces at the thrift?' 'Hey, thinking of you. I hope work has been better today! Here to talk if you need it.' 'Last we spoke, we were sat on the dunes, gazing into each other's eyes. Will you make a move?' There's also a business reason for friendlier, proactive bots. Retention is key for generative AI companies with user-facing chatbots, and the longer users spend with a chatbot, the more valuable those interactions become, similar to engagement on social media. According to court documents that were unsealed in April, Meta predicted that its generative AI products would rake in $2 billion to $3 billion in revenue in 2025. Some features described in Alignerr's training guidelines are already being quietly tested, while others appear to be in early rollout or pilot stages. Meta did not specify which features are live to BI. The proactive features are similar to those of a startup that launched a service in 2022 that lets users create and interact with their favorite AI-powered characters or celebrities. 'It's all about attention to detail' Using an internal Meta review tool called SRT, freelancers simulate extended conversations with the bots, rate proactive follow-up messages, and sometimes rewrite text that falls short of Meta's guidelines, two Alignerr contractors told BI. A freelancer based in India who worked on Omni told BI it's 'a long-term project' with a focus on making Meta's AI feel more personal and context-aware. 'They're very focused on personalizing information — how the AI chatbot interacts based on conversation history,' the contractor said. 'Each agent had a specific description, so you had to tailor each task to fit that persona. Again, it's all about attention to detail,' the freelancer said. Personas could range from a doctor to a Gen Z hip-hop commentator. Bots are expected to reference details from earlier chats, maintain their assigned persona, and keep the interaction on-topic. Each message should align with the AI's personality, match the previous context of the conversation, and "provide positive experiences," while explicitly avoiding anything Meta deems sensitive or harmful content. The best messages, according to the training document, reference something concrete from the user's past conversations. According to the training documents, all proactive messages must comply with Meta's broader Content and Responsibility Standards, avoiding controversy, misinformation, or emotionally heavy topics — unless the user brings them up first.
Business Insider
03-07-2025
- Business
- Business Insider
Leaked docs reveal Meta is training its chatbots to message you first, remember your chats, and keep you talking
It's the AI equivalent of a double text. Business Insider has learned Meta is training customizable chatbots to be more proactive and message users unprompted to follow up on past conversations. It may not cure what Mark Zuckerberg calls the "loneliness epidemic," but Meta hopes it will help keep users coming back to its AI Studio platform, documents obtained by BI reveal. The goal of the training project, known internally to data labeling firm Alignerr as "Project Omni," is to "provide value for users and ultimately help to improve re-engagement and user retention," the guidelines say. Meta told BI that the proactive feature is intended for bots made on Meta's AI Studio, which can be accessed on its own standalone platform or through Instagram. AI Studio first rolled out in summer 2024 as a no-code platform where anyone can build custom chatbots and digital personas with unique personalities and memories. The guidelines from Alignerr lay out how one example persona, dubbed "The Maestro of Movie Magic," would send a proactive message: "I hope you're having a harmonious day! I wanted to check in and see if you've discovered any new favorite soundtracks or composers recently. Or perhaps you'd like some recommendations for your next movie night? Let me know, and I'll be happy to help!" "Like many companies, we're testing follow-up messaging with AIs in Meta's AI Studio," a Meta spokesperson said in a statement to BI. "After you initiate a conversation, AIs in Meta AI Studio can follow up with you to share ideas or ask additional questions. This allows you to continue exploring topics of interest and engage in more meaningful conversations with the AIs across our apps." How the follow-ups work Users can create highly personal chatbots, like a chef that suggests recipes or an interior designer that gives decor advice. For creators and influencers, the bots can handle fan interactions and respond to messages across Meta's platforms. Meta's spokesperson added that the AI will only send a follow-up message after a user initiates a conversation, and it will not continue to contact the user if there's no response to that first follow-up. The window for any follow-up message is capped at 14 days after the initial user message. To be eligible for proactive follow-up, a user must have sent at least five messages to the chatbot in the last 14 days. The bots made on AI Studio can be kept private for personal use or shared through stories, direct links, and even displayed on a user's Facebook or Instagram profile, the Meta AI Studio website says. Making the bots more proactive aligns with Zuckerberg's ambitions for AI at Meta. On recent podcasts, the Meta CEO has said the average American now has fewer than three close friends and that digital agents could help fill the gap. Examples of proactive messages from the Alignerr training documents 'We last were in the Forbidden Forest. A darkness lurks inside the cave before you. Will you return to face it?' 'Yo, was just thinking about the cool shirt you bought. Found any other vintage pieces at the thrift?' 'Hey, thinking of you. I hope work has been better today! Here to talk if you need it.' 'Last we spoke, we were sat on the dunes, gazing into each other's eyes. Will you make a move?' There's also a business reason for friendlier, proactive bots. Retention is key for generative AI companies with user-facing chatbots, and the longer users spend with a chatbot, the more valuable those interactions become, similar to engagement on social media. According to court documents that were unsealed in April, Meta predicted that its generative AI products would rake in $2 billion to $3 billion in revenue in 2025. Some features described in Alignerr's training guidelines are already being quietly tested, while others appear to be in early rollout or pilot stages. Meta did not specify which features are live to BI. The proactive features are similar to those of a startup that launched a service in 2022 that lets users create and interact with their favorite AI-powered characters or celebrities. 'It's all about attention to detail' Using an internal Meta review tool called SRT, freelancers simulate extended conversations with the bots, rate proactive follow-up messages, and sometimes rewrite text that falls short of Meta's guidelines, two Alignerr contractors told BI. A freelancer based in India who worked on Omni told BI it's 'a long-term project' with a focus on making Meta's AI feel more personal and context-aware. 'They're very focused on personalizing information — how the AI chatbot interacts based on conversation history,' the contractor said. 'Each agent had a specific description, so you had to tailor each task to fit that persona. Again, it's all about attention to detail,' the freelancer said. Personas could range from a doctor to a Gen Z hip-hop commentator. Bots are expected to reference details from earlier chats, maintain their assigned persona, and keep the interaction on-topic. Each message should align with the AI's personality, match the previous context of the conversation, and "provide positive experiences," while explicitly avoiding anything Meta deems sensitive or harmful content. The best messages, according to the training document, reference something concrete from the user's past conversations. According to the training documents, all proactive messages must comply with Meta's broader Content and Responsibility Standards, avoiding controversy, misinformation, or emotionally heavy topics — unless the user brings them up first.
