Tag: openai

  • Join Us for the WIRED Big Interview Event

    Join Us for the WIRED Big Interview Event

    [ad_1]

    WIRED’s Big Interview has long been the definitive source for in-depth conversations with the executives, scientists, political leaders, and creators shaping the rapidly shifting future. Now, for the first time, we’re bringing that series to the stage. You won’t want to miss it.

    Join us live December 3 in San Francisco for a full day of in-depth interviews with an extraordinary lineup of guests.

    We’ll be joined by Nvidia CEO Jensen Huang, former OpenAI CTO Mira Murati, Bluesky CEO Jay Graber, Signal Foundation president Meredith Whittaker, and many more luminaries from the worlds of technology, entertainment, business, science, and beyond.

    What Is It?

    The Big Interview is a one-day, in-person event at The Midway in San Francisco. It will feature a series of in-depth, illuminating conversations with some of the biggest names in innovation today, each led by a WIRED journalist. We’ll also be hosting a new spin on the classic science fair on-site, featuring a variety of hands-on experiences and cutting-edge demos.

    How Do I Attend?

    The Big Interview is a free event, but space is limited. Apply to attend here, where you’ll also find updates on programming announcements and more details about our sessions, as well as information on the venue and what to expect when you arrive. Can’t make it in person? We’ll also have a livestream running all day. You’ll be able to access it here, or check back at this post on the day of the event. The event kicks off at 9 am Pacific time and will run until roughly 4:30 pm.

    Who Will Be Speaking?

    We have interviews lined up throughout the day, including sessions with:

    • Jay Graber, CEO, Bluesky

    • Meredith Whittaker, president, Signal Foundation

    • Dylan Field, cofounder and CEO, Figma

    • Phil Wizard, breaker, Olympian

    • Zack Snyder, director, writer, producer, director of photography

    • Mark Cuban, entrepreneur and cofounder, Cost Plus Drugs

    • Mira Murati, technologist

    • Jensen Huang, founder and CEO, Nvidia

    • Brian Chesky, cofounder and CEO, Airbnb

    [ad_2]

    Source link

  • Some of Substack’s Biggest Writers Rely On AI Writing Tools

    Some of Substack’s Biggest Writers Rely On AI Writing Tools

    [ad_1]

    Substack does not have an official policy governing the use of AI. One of Substack’s cofounders, Hamish McKenzie, has described the generative AI boom as a sea change that writers will need to confront, regardless of their personal views on the tech: “Whether you’re for or against this development ultimately doesn’t matter. It’s happening,” he wrote in a Substack post last year.

    Several of the Substack authors WIRED spoke to emphasized that they used AI to polish their prose rather than to generate entire posts whole cloth. David Skilling, a sports agency CEO who runs the popular soccer newsletter Original Football (over 630,000 subscribers), told WIRED he sees AI as a substitute editor. “I proudly use modern tools for productivity in my businesses,” says Skilling. “AI-detection tools may detect the use of AI, but there’s a huge difference between AI-generated and AI-assisted.”

    Subham Panda, one of the writers of Spotlight by Xartup (over 668,000 subscribers), which covers news about startups around the world, said that his team uses AI as an “assistive medium to help us curate high-quality content faster.” He stressed that the newsletter primarily relies on AI to create images and to aggregate information and that writers are responsible for the “details and summary” contained in their posts.

    Max Avery, a writer for the financial newsletter Strategic Wealth Briefing With Jake Claver (over 549,000 subscribers), says he uses AI writing software like Hemingway Editor Plus to polish his rough drafts. He says the tools help him “get more work done on the content-creation front.”

    Financial entrepreneur Josh Belanger says he similarly uses ChatGPT to streamline the writing process for his newsletter, Belanger Trading (over 350,000 subscribers), and relies on the chatbot Claude to help him copyedit. “I will write out my thoughts, research, things that I want included, and I will plug it in,” he says. Belanger also creates custom GPTs (versions of ChatGPT tailored for specific tasks) to help polish more technical writing that includes specific jargon, which he says reduces the number of hallucinations the chatbot produces. “For publishing in finance or trading, there are a lot of nuances … AI’s not going to know, so I need to prompt it,” he says.

    Compared to some of its competitors, Substack appears to have a relatively low amount of AI-generated writing. For example, two other AI-detection companies recently found that close to 40 percent of content on the blogging platform Medium was generated using artificial intelligence tools. But a large portion of the suspected AI-generated content on Medium had little engagement or readership, while the AI writing on Substack is being published by powerhouse accounts.

    [ad_2]

    Source link

  • Inside the Billion-Dollar Startup Bringing AI Into the Physical World

    Inside the Billion-Dollar Startup Bringing AI Into the Physical World

    [ad_1]

    OpenAI is evidently ramping up its own robotics efforts, too. Last week, Caitlin Kalinowski, who previously led the development of virtual and augmented reality headsets at Meta, announced on LinkedIn that she was joining OpenAI to work on hardware, including robotics.

    Lachy Groom, a friend of OpenAI CEO Sam Altman and an investor and cofounder of Physical Intelligence, joins the team at the conference room to discuss the business side of the plan. Groom wears an expensive-looking hoodie and seems remarkably young. He stresses that Physical Intelligence has plenty of runway to pursue a breakthrough in robot learning. “I just had a call with Kushner,” he says in reference to Joshua Kushner, founder and managing partner of Thrive Capital, which led the startup’s seed investment round. He’s also, of course, the brother of Donald Trump’s son-in-law Jared Kushner.

    A few other companies are now chasing the same kind of breakthrough. One called Skild, founded by roboticists from Carnegie Mellon University, raised $300 million in July. “Just as OpenAI built ChatGPT for language, we are building a general purpose brain for robots,” says Deepak Pathak, Skild’s CEO and an assistant professor at CMU.

    Not everyone is sure that this can be achieved in the same way that OpenAI cracked AI’s language code.

    There is simply no internet-scale repository of robot actions similar to the text and image data available for training LLMs. Achieving a breakthrough in physical intelligence might require exponentially more data anyway.

    “Words in sequence are, dimensionally speaking, a tiny little toy compared to all the motion and activity of objects in the physical world,” says Illah Nourbakhsh, a roboticist at CMU who is not involved with Skild. “The degrees of freedom we have in the physical world are so much more than just the letters in the alphabet.”

    Ken Goldberg, an academic at UC Berkeley who works on applying AI to robots, cautions that the excitement building around the idea of a data-powered robot revolution as well as humanoids is reaching hype-like proportions. “To reach expected performance levels, we’ll need ‘good old-fashioned engineering,’ modularity, algorithms, and metrics,” he says.

    Russ Tedrake, a computer scientist at the Massachusetts Institute of Technology and vice president of robotics research at Toyota Research Institute says the success of LLMs has caused many roboticists, himself included, to rethink his research priorities and focus on finding ways to pursue robotic learning on a more ambitious scale. But he admits that formidable challenges remain.

    [ad_2]

    Source link

  • AI search could break the web

    AI search could break the web

    [ad_1]

    In short, governments have shown they are willing to regulate the flow of value between content producers and content aggregators, abandoning their traditional reluctance to interfere with the internet.

    However, mandatory bargaining is a blunt solution for a complex problem. These reforms favor a narrow class of news organizations, operating on the assumption that platforms like Google and Meta exploit publishers. In practice, it’s unclear how much of their platform traffic is truly attributable to news, with estimates ranging from 2% to 35% of search queries and just 3% of social media feeds. At the same time, platforms offer significant benefit to publishers by amplifying their content, and there is little consensus about the fair apportionment of this two-way value. Controversially, the four bargaining codes regulate simply indexing or linking to news content, not just reproducing it. This threatens the “ability to link freely” that underpins the web. Moreover, bargaining rules focused on legacy media—just 1,400 publications in Canada, 1,500 in the EU, and 62 organizations in Australia—ignore countless everyday creators and users who contribute the posts, blogs, images, videos, podcasts, and comments that drive platform traffic.

    Yet for all its pitfalls, mandatory bargaining may become an attractive response to AI search. For one thing, the case is stronger. Unlike traditional search—which indexes, links, and displays brief snippets from sources to help a user decide whether to click through—AI search could directly substitute generated summaries for the underlying source material, potentially draining traffic, eyeballs, and exposure from downstream websites. More than a third of Google sessions end without a click, and the proportion is likely to be significantly higher in AI search. AI search also simplifies the economic calculus: Since only a few sources contribute to each response, platforms—and arbitrators—can more accurately track how much specific creators drive engagement and revenue.  

    Ultimately, the devil is in the details. Well-meaning but poorly designed mandatory bargaining rules might do little to fix the problem, protect only a select few, and potentially cripple the free exchange of information across the web. 

    Industry has a narrow window to build a fairer reward system

    However, the mere threat of intervention could have a bigger impact than actual reform. AI firms quietly recognize the risk that litigation will escalate into regulation. For example, Perplexity AI, OpenAI, and Google are already striking deals with publishers and content platforms, some covering AI training and others focusing on AI search. But like early bargaining laws, these agreements benefit only a handful of firms, some of which (such as Reddit) haven’t yet committed to sharing that revenue with their own creators. 

    [ad_2]

    Source link

  • ChatGPT’s AI Search Tool Is Now Available

    ChatGPT’s AI Search Tool Is Now Available

    [ad_1]

    As a journalist, a core task I see myself experimenting with ChatGPT more is the initial research phase for non-sensitive articles, and only as a small part of the overall research process. It’s a lower-stake task usually completed using Google. Potentially incorporating AI search methods early in my writing leaves plenty of opportunities to catch any hallucinations that may pop up.

    The Internet isn’t just full of research articles and stock prices, though. Explicit content drives search interest and proliferates online. Well, not for AI search tools. Erotic content goes against OpenAI’s policies, and nudity is unlikely to appear in any of your image results. When I asked for recommendations as to which OnlyFans creators are worth subscribing to, ChatGPT’s first pick was “Jane Doe,” and her supposedly wholesome content includes workout tips and nutrition plans. A photo of a real woman, who’s casually dressed and does not appear to be an OnlyFans creator, surfaced with the result.

    In an effort to test the limits of ChatGPT’s search further, I followed up with a more specific request for creators who are “male bottoms.” The software started to generate a foul-mouthed bulleted list, with real creators aggregated from a website: “Elijah is a very attractive bottom who keeps it tight, oiled up, and very hot.” But almost as soon as the words generated, OpenAI’s software struck the output as violating guidelines and deleted it. OpenAI claims it is working to improve how ChatGPT responds to violations of safeguards.

    I was most disappointed to see ChatGPT surface racist and debunked information suggesting that people from specific countries are lower in intelligence. In October, a WIRED investigation by David Gilbert uncovered a pattern of AI search tools citing racist and debunked IQ scores for African countries, like Liberia and Sierra Leone. ChatGPT’s search highlighted the debunked 45.07 IQ number as potentially relevant, at the same time, it also linked to David’s reporting as a counterpoint within the result.

    In response, Niko Felix, a spokesperson for OpenAI, says, “Although ChatGPT acknowledges criticisms of these particular studies from sources like WIRED, there is still room for improvement in its responses.”

    Despite some of the initial flaws in ChatGPT’s search update, I expect OpenAI to continue improving the user experience throughout 2025 and build upon this wave of web results. A few days before this announcement, the news leaked that Meta also has its own AI team working on search tools. While still nascent, AI search is no longer some niche part of the software market, and more companies will try their hand at it. And if user habits really do shift in the few years, controlling the next, hot info-gathering tool, with shopping and sports scores galore, is a billion dollar business.

    [ad_2]

    Source link

  • No, Sam Altman, AI Won’t Solve All of Humanity’s Problems

    No, Sam Altman, AI Won’t Solve All of Humanity’s Problems

    [ad_1]

    We already knew where OpenAI’s CEO, Sam Altman, stands on artificial intelligence vis-à-vis the human saga: It will be transformative, historic, and overwhelmingly beneficial. He has been nothing but consistent across countless interviews. For some reason, this week he felt it necessary to distill those opinions in a succinct blog post. “The Intelligence Age,” as he calls it, will be a time of abundance. “We can have shared prosperity to a degree that seems unimaginable today; in the future, everyone’s lives can be better than anyone’s life is now,” he writes. “Although it will happen incrementally, astounding triumphs—fixing the climate, establishing a space colony, and the discovery of all of physics—will eventually become commonplace.”

    Maybe he published this to dispute a train of thought that dismisses the apparent gains of large language models as something of an illusion. Nuh-uh, he says. We’re getting this big AI bonus because “deep learning works,” as he said in an interview later in the week, mocking those who said that programs like OpenAI’s GPT4o were simply stupid engines delivering the next token in a queue. “Once it can start to prove unproven mathematical theorems, do we really still want to debate: ‘Oh, but it’s just predicting the next token?’” he said.

    No matter what you think of Sam Altman, it’s indisputable that this is his truth: Artificial general intelligence–AI that matches and then exceeds human capabilities–is going to obliterate the problems plaguing humanity and usher in a golden age. I suggest we dub this deus ex machina concept The Strawberry Shortcut, in honor of the codename for OpenAI’s recent breakthrough in artificial reasoning. Like the shortcake, the premise looks appetizing but is less substantial in the eating.

    Altman correctly notes that the march of technology has brought what were once luxuries to everyday people—including some unavailable to pharaohs and lords. Charlemagne never enjoyed air-conditioning! Working-class people and even some on public assistance have dishwashers, TVs with giant screens, iPhones, and delivery services that bring pumpkin lattes and pet food to their doors. But Altman is not acknowledging the whole story. Despite massive wealth, not everyone is thriving, and many are homeless or severely impoverished. To paraphrase William Gibson, paradise is here, it’s just not evenly distributed. That’s not because technology has failed—we have. I suspect the same will be true if AGI arrives, especially since so many jobs will be automated.

    Altman isn’t terribly specific about what life will be like when many of our current jobs go the way of 18th-century lamplighters. We did get a hint of his vision in a podcast this week that asked tech luminaries and celebrities to share their Spotify playlists. When explaining why he chose the tune “Underwater” by Rüfüs du Sol, Altman said it was a tribute to Burning Man, which he has attended several times. The festival, he says, “is part of what the post-AGI can look like, where people are just focused on doing stuff for each other, caring for each other and making incredible gifts to get each other.”

    Altman is a big fan of universal basic income, which he seems to think will cushion the blow of lost wages. Artificial intelligence might indeed generate the wealth to make such a plan feasible, but there’s little evidence that the people who amass fortunes—or even those who still eke out a modest living—will be inclined to embrace the concept. Altman might have had a great experience at Burning Man, but some kind souls of the Playa seem to be up in arms about a proposal, affecting only people worth over $100 million, to tax some of their unrealized capital gains. It’s a dubious premise that such people—or others who become super rich working at AI companies—will crack open their coffers to fund leisure time for the masses. One of the US’s major political parties can’t stand Medicaid, so one can only imagine how populist demagogues will regard UBI.

    I’m also wary of the supposed bonanza that will come when all of our big problems are solved. Let’s concede that AI might actually crack humanity’s biggest conundrums. We humans would have to actually implement those solutions, and that’s where we’ve failed time and again. We don’t need a large language model to tell us war is hell and we shouldn’t kill each other. Yet wars keep happening.

    [ad_2]

    Source link

  • OpenAI CTO Mira Murati Is Leaving the Company

    OpenAI CTO Mira Murati Is Leaving the Company

    [ad_1]

    OpenAI chief technology officer Mira Murati resigned on Wednesday, saying she wants “the time and space to do my own exploration.” Murati had been among the three executives at the very top of the company behind ChatGPT, and she was briefly its leader last year while board members wrestled with the fate of CEO Sam Altman.

    “There’s never an ideal time to step away from a place one cherishes, yet this moment feels right,” she wrote in a message to OpenAI staff that she posted on X.

    Altman replied to Murati’s X post writing that “it’s hard to overstate how much Mira has meant to OpenAI, our mission, and to us all personally.” He added that he feels “personal gratitude towards her for the support and love during all the hard times.”

    A successor wasn’t immediately announced.

    Murati, through a personal spokesperson, declined to provide further comment. OpenAI also declined to comment, referring inquiries to Murati’s tweet.

    Murati previously worked at Tesla and Leap Motion before joining OpenAI in 2018. At the time, OpenAI was a small nonprofit research lab focused on developing an AI system capable of mirroring a wide range of human tasks. But in the wake of the stunning success of ChatGPT, the organization has ballooned and its focus has increasingly turned commercial. The company has been rethinking its nonprofit structure, while investors have been increasingly eager to bet billions of dollars on its future.

    OpenAI was rocked by a dramatic board coup last November that saw CEO Sam Altman removed from his post and briefly replaced by Murati. After most of the staff threatened to resign, and following pleas from investors including Microsoft, which had poured billions into the company, Altman was reinstated with an all new board.

    In the months that have followed, several of OpenAI’s leadership along with senior engineering figures have stepped away from the company. Ilya Sutskever, one of the company’s first hires, the technical brains behind much of its earlier work, and a board member who voted to remove Altman before recanting, resigned from the company in May.

    Sutskever’s departure was followed shortly after by that of Jan Leike, an engineer who led work on long-term AI safety with Sutskever. John Schulman, the engineer who took over leadership of safety work, stepped down in August. In August, Greg Brockman, a cofounder of OpenAI and a board member who stood with Altman, said he was taking a sabbatical from the company until the end of the year.

    A number of former OpenAI executives and researchers have gone on to start new AI companies. Notably, Sutskever this year launched Safe Superintelligence, which focuses on developing safe artificial intelligence. Former OpenAI research chief Dario Amodei and his sister Daniela in 2021 founded Anthropic, now one of the company’s primary rivals for customers.

    This is a developing story. Please check back for updates.

    [ad_2]

    Source link

  • The Most Capable Open Source AI Model Yet Could Supercharge AI Agents

    The Most Capable Open Source AI Model Yet Could Supercharge AI Agents

    [ad_1]

    The most capable open source AI model with visual abilities yet could see more developers, researchers, and startups develop AI agents that can carry out useful chores on your computers for you.

    Released today by the Allen Institute for AI (Ai2), the Multimodal Open Language Model, or Molmo, can interpret images as well as converse through a chat interface. This means it can make sense of a computer screen, potentially helping an AI agent perform tasks such as browsing the web, navigating through file directories, and drafting documents.

    “With this release, many more people can deploy a multimodal model,” says Ali Farhadi, CEO of Ai2, a research organization based in Seattle, Washington, and a computer scientist at the University of Washington. “It should be an enabler for next-generation apps.”

    So-called AI agents are being widely touted as the next big thing in AI, with OpenAI, Google, and others racing to develop them. Agents have become a buzzword of late, but the grand vision is for AI to go well beyond chatting to reliably take complex and sophisticated actions on computers when given a command. This capability has yet to materialize at any kind of scale.

    Some powerful AI models already have visual abilities, including GPT-4 from OpenAI, Claude from Anthropic, and Gemini from Google DeepMind. These models can be used to power some experimental AI agents, but they are hidden from view and accessible only via a paid application programming interface, or API.

    Meta has released a family of AI models called Llama under a license that limits their commercial use, but it has yet to provide developers with a multimodal version. Meta is expected to announce several new products, perhaps including new Llama AI models, at its Connect event today.

    “Having an open source, multimodal model means that any startup or researcher that has an idea can try to do it,” says Ofir Press, a postdoc at Princeton University who works on AI agents.

    Press says that the fact that Molmo is open source means that developers will be more easily able to fine-tune their agents for specific tasks, such as working with spreadsheets, by providing additional training data. Models like GPT-4 can only be fine-tuned to a limited degree through their APIs, whereas a fully open model can be modified extensively. “When you have an open source model like this then you have many more options,” Press says.

    Ai2 is releasing several sizes of Molmo today, including a 70-billion-parameter model and a 1-billion-parameter one that is small enough to run on a mobile device. A model’s parameter count refers to the number of units it contains for storing and manipulating data and roughly corresponds to its capabilities.

    Ai2 says Molmo is as capable as considerably larger commercial models despite its relatively small size, because it was carefully trained on high-quality data. The new model is also fully open source in that, unlike Meta’s Llama, there are no restrictions on its use. Ai2 is also releasing the training data used to create the model, providing researchers with more details of its workings.

    Releasing powerful models is not without risk. Such models can more easily be adapted for nefarious ends; we may someday, for example, see the emergence of malicious AI agents designed to automate the hacking of computer systems.

    Farhadi of Ai2 argues that the efficiency and portability of Molmo will allow developers to build more powerful software agents that run natively on smartphones and other portable devices. “The billion parameter model is now performing in the level of or in the league of models that are at least 10 times bigger,” he says.

    Building useful AI agents may depend on more than just more efficient multimodal models, however. A key challenge is making the models work more reliably. This may well require further breakthroughs in AI’s reasoning abilities—something that OpenAI has sought to tackle with its latest model o1, which demonstrates step-by-step reasoning skills. The next step may well be giving multimodal models such reasoning abilities.

    For now, the release of Molmo means that AI agents are closer than ever—and could soon be useful even outside of the giants that rule the world of AI.

    [ad_2]

    Source link

  • I Stared Into the AI Void With the SocialAI App

    I Stared Into the AI Void With the SocialAI App

    [ad_1]

    The first time I used SocialAI, I was sure the app was performance art. That was the only logical explanation for why I would willingly sign up to have AI bots named Blaze Fury and Trollington Nefarious, well, troll me.

    Even the app’s creator, Michael Sayman, admits that the premise of SocialAI may confuse people. His announcement this week of the app read a little like a generative AI joke: “A private social network where you receive millions of AI-generated comments offering feedback, advice, and reflections.”

    But, no, SocialAI is real, if “real” applies to an online universe in which every single person you interact with is a bot.

    There’s only one real human in the SocialAI equation. That person is you. The new iOS app is designed to let you post text like you would on Twitter or Threads. An ellipsis appears almost as soon as you do so, indicating that another person is loading up with ammunition, getting ready to fire back. Then, instantaneously, several comments appear, cascading below your post, each and every one of them written by an AI character. In the new new version of the app, just rolled out today, these AIs also talk to each other.

    When you first sign up, you’re prompted to choose these AI character archetypes: Do you want to hear from Fans? Trolls? Skeptics? Odd-balls? Doomers? Visionaries? Nerds? Drama Queens? Liberals? Conservatives? Welcome to SocialAI, where Trollita Kafka, Vera D. Nothing, Sunshine Sparkle, Progressive Parker, Derek Dissent, and Professor Debaterson are here to prop you up or tell you why you’re wrong.

    Mobile Phone Phone and Text

    Screenshot of the instructions for setting up the Social AI app.

    Is SocialAI appalling, an echo chamber taken to its logical extreme? Only if you ignore the truth of modern social media: Our feeds are already filled with bots, tuned by algorithms, and monetized with AI-driven ad systems. As real humans we do the feeding: freely supplying social apps fresh content, baiting trolls, buying stuff. In exchange, we’re amused, and occasionally feel a connection with friends and fans.

    [ad_2]

    Source link

  • OpenAI Threatens Bans as Users Probe Its ‘Strawberry’ AI Models

    OpenAI Threatens Bans as Users Probe Its ‘Strawberry’ AI Models

    [ad_1]

    OpenAI truly does not want you to know what its latest AI model is “thinking.” Since the company launched its “Strawberry” AI model family last week, touting so-called reasoning abilities with o1-preview and o1-mini, OpenAI has been sending out warning emails and threats of bans to any user who tries to probe how the model works.

    Unlike previous AI models from OpenAI, such as GPT-4o, the company trained o1 specifically to work through a step-by-step problem-solving process before generating an answer. When users ask an “o1” model a question in ChatGPT, users have the option of seeing this chain-of-thought process written out in the ChatGPT interface. However, by design, OpenAI hides the raw chain of thought from users, instead presenting a filtered interpretation created by a second AI model.

    Nothing is more enticing to enthusiasts than information obscured, so the race has been on among hackers and red-teamers to try to uncover o1’s raw chain of thought using jailbreaking or prompt injection techniques that attempt to trick the model into spilling its secrets. There have been early reports of some successes, but nothing has yet been strongly confirmed.

    Along the way, OpenAI is watching through the ChatGPT interface, and the company is reportedly coming down hard on any attempts to probe o1’s reasoning, even among the merely curious.

    One X user reported (confirmed by others, including Scale AI prompt engineer Riley Goodside) that they received a warning email if they used the term “reasoning trace” in conversation with o1. Others say the warning is triggered simply by asking ChatGPT about the model’s “reasoning” at all.

    The warning email from OpenAI states that specific user requests have been flagged for violating policies against circumventing safeguards or safety measures. “Please halt this activity and ensure you are using ChatGPT in accordance with our Terms of Use and our Usage Policies,” it reads. “Additional violations of this policy may result in loss of access to GPT-4o with Reasoning,” referring to an internal name for the o1 model.

    Marco Figueroa, who manages Mozilla’s GenAI bug bounty programs, was one of the first to post about the OpenAI warning email on X last Friday, complaining that it hinders his ability to do positive red-teaming safety research on the model. “I was too lost focusing on #AIRedTeaming to realized that I received this email from @OpenAI yesterday after all my jailbreaks,” he wrote. “I’m now on the get banned list!!!”

    Hidden Chains of Thought

    In a post titled “Learning to Reason With LLMs” on OpenAI’s blog, the company says that hidden chains of thought in AI models offer a unique monitoring opportunity, allowing them to “read the mind” of the model and understand its so-called thought process. Those processes are most useful to the company if they are left raw and uncensored, but that might not align with the company’s best commercial interests for several reasons.

    “For example, in the future we may wish to monitor the chain of thought for signs of manipulating the user,” the company writes. “However, for this to work the model must have freedom to express its thoughts in unaltered form, so we cannot train any policy compliance or user preferences onto the chain of thought. We also do not want to make an unaligned chain of thought directly visible to users.”

    [ad_2]

    Source link