ReportWire

Tag: fast forward

  • Could This Be the Start of Amazon’s Next Robot Revolution?

    Could This Be the Start of Amazon’s Next Robot Revolution?

    In 2012, Amazon quietly acquired a robotics startup called Kiva Systems, a move that dramatically improved the efficiency of its ecommerce operations and kickstarted a wider revolution in warehouse automation.

    Last week, the ecommerce giant announced another deal that could prove similarly profound, agreeing to hire the founders of Covariant, a startup that has been testing ways for AI to automate more of the picking and handling of a wide range of physical objects.

    Covariant may have found it challenging to commercialize AI-infused industrial robots given the high costs and sharp competition involved; the deal, which will also see Amazon license Covariant’s models and data, could bring about another revolution in ecommerce—one that might prove hard for any competitor to match given Amazon’s vast operational scale and data trove.

    The deal is also an example of a Big Tech company acquiring core talent and expertise from an AI startup without actually buying the company outright. Amazon came to a similar agreement with the startup Adept in June. In March, Microsoft struck a deal with Inflection, and in August, Google hired the founders of Character AI.

    Back in the aughts, Kiva developed a way to move products through warehouses by having squat robots lift and carry stocked shelves over to human pickers—a trick that meant workers no longer needed to walk miles every day to find different items. Kiva’s mobile bots were similar to those employed in manufacturing, and the company used clever algorithms to coordinate the movement of thousands of bots in the same physical space.

    Amazon’s mobile robot army grew from around 10,000 in 2013 to 750,000 by 2023, and the sheer scale of the company’s operations meant that it could deliver millions of items faster and cheaper than anyone else.

    As WIRED revealed last year, Amazon has in recent years developed new robotic systems that rely on machine learning to do things like perceive, grab, and sort packed boxes. Again, Amazon is leveraging scale to its advantage, with the training data being gathered as items flow through its facilities helping to improve the performance of different algorithms. The effort has already led to further automation of the work that had previously been done by human workers at some fulfillment centers.

    The one chore that remains stubbornly difficult to mechanize, however, is the physical grasping of products. It requires adaptability to account for things like friction and slippage, and robots will inevitably be confronted with unfamiliar and awkward items among Amazon’s vast inventory.

    Covariant has spent the past few years developing AI algorithms with a more general ability to handle a range of items more reliably. The company was founded in 2020 by Pieter Abbeel, a professor at UC Berkeley who has done pioneering work on applying machine learning to robotics, along with several of his students, including Peter Chen, who became Covariant’s CEO, and Rocky Duan, the company’s CTO. This week’s deal will see all three of them, along with several research scientists at the startup, join Amazon.

    “Covariant’s models will be used to power some of the robotic manipulation systems across our fulfillment network,” Alexandra Miller, an Amazon spokesperson, tells WIRED. The tech giant declined to reveal financial details of the deal.

    Abbeel was an early employee at OpenAI, and his company has taken inspiration from the story of ChatGPT’s success. In March, Covariant demonstrated a chat interface for its robot and said it had developed a foundation model for robotic grasping, meaning an algorithm designed to become

    Will Knight

    Source link

  • OpenAI Warns Users Could Become Emotionally Hooked on Its Voice Mode

    OpenAI Warns Users Could Become Emotionally Hooked on Its Voice Mode

    In late July, OpenAI began rolling out an eerily humanlike voice interface for ChatGPT. In a safety analysis released today, the company acknowledges that this anthropomorphic voice may lure some users into becoming emotionally attached to their chatbot.

    The warnings are included in a “system card” for GPT-4o, a technical document that lays out what the company believes are the risks associated with the model, plus details surrounding safety testing and the mitigation efforts the company’s taking to reduce potential risk.

    OpenAI has faced scrutiny in recent months after a number of employees working on AI’s long-term risks quit the company. Some subsequently accused OpenAI of taking unnecessary chances and muzzling dissenters in its race to commercialize AI. Revealing more details of OpenAI’s safety regime may help mitigate the criticism and reassure the public that the company takes the issue seriously.

    The risks explored in the new system card are wide-ranging, and include the potential for GPT-4o to amplify societal biases, spread disinformation, and aid in the development of chemical or biological weapons. It also discloses details of testing designed to ensure that AI models won’t try to break free of their controls, deceive people, or scheme catastrophic plans.

    Some outside experts commend OpenAI for its transparency but say it could go further.

    Lucie-Aimée Kaffee, an applied policy researcher at Hugging Face, a company that hosts AI tools, notes that OpenAI’s system card for GPT-4o does not include extensive details on the model’s training data or who owns that data. “The question of consent in creating such a large dataset spanning multiple modalities, including text, image, and speech, needs to be addressed,” Kaffee says.

    Others note that risks could change as tools are used in the wild. “Their internal review should only be the first piece of ensuring AI safety,” says Neil Thompson, a professor at MIT who studies AI risk assessments. “Many risks only manifest when AI is used in the real world. It is important that these other risks are cataloged and evaluated as new models emerge.”

    The new system card highlights how rapidly AI risks are evolving with the development of powerful new features such as OpenAI’s voice interface. In May, when the company unveiled its voice mode, which can respond swiftly and handle interruptions in a natural back and forth, many users noticed it appeared overly flirtatious in demos. The company later faced criticism from the actress Scarlett Johansson, who accused it of copying her style of speech.

    A section of the system card titled “Anthropomorphization and Emotional Reliance” explores problems that arise when users perceive AI in human terms, something apparently exacerbated by the humanlike voice mode. During the red teaming, or stress testing, of GPT-4o, for instance, OpenAI researchers noticed instances of speech from users that conveyed a sense of emotional connection with the model. For example, people used language such as “This is our last day together.”

    Anthropomorphism might cause users to place more trust in the output of a model when it “hallucinates” incorrect information, OpenAI says. Over time, it might even affect users’ relationships with other people. “Users might form social relationships with the AI, reducing their need for human interaction—potentially benefiting lonely individuals but possibly affecting healthy relationships,” the document says.

    Joaquin Quiñonero Candela, head of preparedness at OpenAI, says that voice mode could evolve into a uniquely powerful interface. He also notes that the kind of emotional effects seen with GPT-4o can be positive—say, by helping those who are lonely or who need to practice social interactions. He adds that the company will study anthropomorphism and the emotional connections closely, including by monitoring how beta testers interact with ChatGPT. “We don’t have results to share at the moment, but it’s on our list of concerns,” he says.

    Will Knight, Reece Rogers

    Source link

  • Google DeepMind’s Game-Playing AI Tackles a Chatbot Blind Spot

    Google DeepMind’s Game-Playing AI Tackles a Chatbot Blind Spot

    Several years before ChatGPT began jibber-jabbering away, Google developed a very different kind of artificial intelligence program called AlphaGo that learned to play the board game Go with superhuman skill through tireless practice.

    Researchers at the company have now published research that combines the abilities of a large language model (the AI behind today’s chatbots) with those of AlphaZero, a successor to AlphaGo also capable of playing chess, to solve very tricky mathematical proofs.

    Their new Frankensteinian creation, dubbed AlphaProof, has demonstrated its prowess by tackling several problems from the 2024 International Math Olympiad (IMO), a prestigious competition for high school students.

    AlphaProof uses the Gemini large language model to convert naturally phrased math questions into a programming language called Lean. This provides the training fodder for a second algorithm to learn, through trial and error, how to find proofs that can be confirmed as correct.

    Earlier this year, Google DeepMind revealed another math algorithm called AlphaGeometry that also combines a language model with a different AI approach. AlphaGeometry uses Gemini to convert geometry problems into a form that can be manipulated and tested by a program that handles geometric elements. Google today also announced a new and improved version of AlphaGeometry.

    The researchers found that their two math programs could provide proofs for IMO puzzles as well as a silver medalist could. Out of six problems total, AlphaProof solved two algebra problems and a number theory one, while AlphaGeometry solved a geometry problem. The programs got one problem in minutes but took up to several days to figure out others. Google DeepMind has not disclosed how much computer power it threw at the problems.

    Google DeepMind calls the approach used for both AlphaProof and AlphaGeometry “neuro-symbolic” because they combine the pure machine learning of an artificial neural network, the technology that underpins most progress in AI of late, with the language of conventional programming.

    “What we’ve seen here is that you can combine the approach that was so successful, and things like AlphaGo, with large language models and produce something that is extremely capable,” says David Silver, the Google DeepMind researcher who led work on AlphaZero. Silver says the techniques demonstrated with AlphaProof should, in theory, extend to other areas of mathematics.

    Indeed, the research raises the prospect of addressing the worst tendencies of large language models by applying logic and reasoning in a more grounded fashion. As miraculous as large language models can be, they often struggle to grasp even basic math or to reason through problems logically.

    In the future, the neural-symbolic method could provide a means for AI systems to turn questions or tasks into a form that can be reasoned over in a way that produces reliable results. OpenAI is also rumored to be working on such a system, codenamed “Strawberry.”

    There is, however, a key limitation with the systems revealed today, as Silver acknowledges. Math solutions are either correct or incorrect, allowing AlphaProof and AlphaGeometry to work their way toward the right answer. Many real-world problems—coming up with the ideal itinerary for a trip, for instance—have many possible solutions, and which one is ideal may be unclear. Silver says the solution for more ambiguous questions may be for a language model to try to determine what constitutes a “right” answer during training. “There’s a spectrum of different things that can be tried,” he says.

    Silver is also careful to note that Google DeepMind won’t be putting human mathematicians out of jobs. “We are aiming to provide a system that can prove anything, but that’s not the end of what mathematicians do,” he says. “A big part of mathematics is to pose problems and find what are the interesting questions to ask. You might think of this as another tool along the lines of a slide rule or calculator or computational tools.”

    Updated 7/25/24 1:25 pm ET: This story has been updated to clarify how many problems AlphaProof and AlphaGeometry solved, and of what type.

    Will Knight

    Source link

  • The AI-Powered Future of Coding Is Near

    The AI-Powered Future of Coding Is Near

    I am by no means a skilled coder, but thanks to a free program called SWE-agent, I was just able to debug and fix a gnarly problem involving a misnamed file within different code repositories on the software-hosting site GitHub.

    I pointed SWE-agent at an issue on GitHub and watched as it went through the code and reasoned about what might be wrong. It correctly determined that the root cause of the bug was a line that pointed to the wrong location for a file, then navigated through the project, located the file, and amended the code so that everything ran properly. It’s the kind of thing that an inexperienced developer (such as myself) might spend hours trying to debug.

    Many coders already use artificial intelligence to write software more quickly. GitHub Copilot was the first integrated developer environment to harness AI, but lots of IDEs will now automatically complete chunks of code when a developer starts typing. You can also ask AI questions about code or have it offer suggestions on how to improve what you’re working on.

    Last summer, John Yang and Carlos Jimenez, two Princeton PhD students, began discussing what it would take for AI to become a real-world software engineer. This led them and others at Princeton to come up with SWE-bench, a set of benchmarks for testing AI tools across a range of coding tasks. After releasing the benchmark in October, the team developed its own tool—SWE-agent—to master these tasks.

    SWE-agent (“SWE” is shorthand for “software engineering”) is one of a number of considerably more powerful AI coding programs that go beyond just writing lines of code and act as so-called software agents, harnessing the tools needed to wrangle, debug, and organize software. The startup Devin went viral with a video demo of one such tool in March.

    Ofir Press, a member of the Princeton team, says that SWE-bench could help OpenAI test the performance and reliability of software agents. “It’s just my opinion, but I think they will release a software agent very soon,” Press says.

    OpenAI declined to comment, but another source with knowledge of the company’s activities, who asked not to be named, told WIRED that “OpenAI is definitely working on coding agents.”

    Just as GitHub Copilot showed that large language models can write code and boost programmers’ productivity, tools like SWE-agent may prove that AI agents can work reliably, starting with building and maintaining code.

    A number of companies are testing agents for software development. At the top of the SWE-bench leaderboard, which measures the score of different coding agents across a variety of tasks, is one from Factory AI, a startup, followed by AutoCodeRover, an open source entry from a team at the National University of Singapore.

    Big players are also wading in. A software-writing tool called Amazon Q is another top performer on SWE-bench. “Software development is a lot more than just typing,” says Deepak Singh, vice president of software development at Amazon Web Services.

    He adds that AWS has used the agent to translate entire software stacks from one programming language to another one. “It’s like having a really smart engineer sitting next to you, writing and building an application with you,” Singh says. “I think that’s pretty transformative.”

    A team at OpenAI recently helped the Princeton crew improve a benchmark for measuring the reliability and efficacy of tools like SWE-agent, suggesting that the company might also be honing agents for writing code or doing other tasks on a computer.

    Singh says that a number of customers are already building complex backend applications using Q. My own experiments with SWE-bench suggest that anyone who codes will soon want to use agents to enhance their programming prowess, or risk being left behind.

    Will Knight

    Source link

  • OpenAI Is Testing Its Powers of Persuasion

    OpenAI Is Testing Its Powers of Persuasion

    This week, Sam Altman, CEO of OpenAI, and Arianna Huffington, founder and CEO of the health company Thrive Global, published an article in Time touting Thrive AI, a startup backed by Thrive and OpenAI’s Startup Fund. The piece suggests that AI could have a huge positive impact on public health by talking people into healthier behavior.

    Altman and Huffington write that Thrive AI is working toward “a fully integrated personal AI coach that offers real-time nudges and recommendations unique to you that allows you to take action on your daily behaviors to improve your health.”

    Their vision puts a positive spin on what may well prove to be one of AI’s sharpest double-edges. AI models are already adept at persuading people, and we don’t know how much more powerful they could become as they advance and gain access to more personal data.

    Aleksander Madry, a professor on sabbatical from the Massachusetts Institute of Technology, leads a team at OpenAI called Preparedness that is working on that very issue.

    “One of the streams of work in Preparedness is persuasion,” Madry told WIRED in a May interview. “Essentially, thinking to what extent you can use these models as a way of persuading people.”

    Madry says he was drawn to join OpenAI by the remarkable potential of language models and because the risks that they pose have barely been studied. “There is literally almost no science,” he says. “That was the impetus for the Preparedness effort.”

    Persuasiveness is a key element in programs like ChatGPT and one of the ingredients that makes such chatbots so compelling. Language models are trained in human writing and dialog that contains countless rhetorical and suasive tricks and techniques. The models are also typically fine-tuned to err toward utterances that users find more compelling.

    Research released in April by Anthropic, a competitor founded by OpenAI exiles, suggests that language models have become better at persuading people as they have grown in size and sophistication. This research involved giving volunteers a statement and then seeing how an AI-generated argument changes their opinion of it.

    OpenAI’s work extends to analyzing AI in conversation with users—something that may unlock greater persuasiveness. Madry says the work is being conducted on consenting volunteers, and declines to reveal the findings to date. But he says the persuasive power of language models runs deep. “As humans we have this ‘weakness’ that if something communicates with us in natural language [we think of it as if] it is a human,” he says, alluding to an anthropomorphism that can make chatbots seem more lifelike and convincing.

    The Time article argues that the potential health benefits of persuasive AI will require strong legal safeguards because the models may have access to so much personal information. “Policymakers need to create a regulatory environment that fosters AI innovation while safeguarding privacy,” Altman and Huffington write.

    This is not all that policymakers will need to consider. It may also be crucial to weigh how increasingly persuasive algorithms could be misused. AI algorithms could enhance the resonance of misinformation or generate particularly compelling phishing scams. They might also be used to advertise products.

    Madry says a key question, yet to be studied by OpenAI or others, is how much more compelling or coercive AI programs that interact with users over long periods of time could prove to be. Already a number of companies offer chatbots that roleplay as romantic partners and other characters. AI girlfriends are increasingly popular—some are even designed to yell at you—but how addictive and persuasive these bots are is largely unknown.

    The excitement and hype generated by ChatGPT following its release in November 2022 saw OpenAI, outside researchers, and many policymakers zero in on the more hypothetical question of whether AI could someday turn against its creators.

    Madry says this risks ignoring the more subtle dangers posed by silver-tongued algorithms. “I worry that they will focus on the wrong questions,” Madry says of the work of policymakers thus far. “That in some sense, everyone says, ‘Oh yeah, we are handling it because we are talking about it,’ when actually we are not talking about the right thing.”

    Will Knight

    Source link

  • We’re Still Waiting for the Next Big Leap in AI

    We’re Still Waiting for the Next Big Leap in AI

    When OpenAI announced GPT-4, its latest large language model, last March, it sent shockwaves through the tech world. It was clearly more capable than anything seen before at chatting, coding, and solving all sorts of thorny problems—including school homework.

    Anthropic, a rival to OpenAI, announced today that it has made its own AI advance that will upgrade chatbots and other use cases. But although the new model is the world’s best by some measures, it’s more of a step forward than a big leap.

    Anthropic’s new model, called Claude 3.5 Sonnet, is an upgrade to its existing Claude 3 family of AI models. It is more adept at solving math, coding, and logic problems as measured by commonly used benchmarks. Anthropic says it is also a lot faster, better understands nuances in language, and even has a better sense of humor.

    That’s no doubt useful to people trying to build apps and services on top of Anthropic’s AI models. But the company’s news is also a reminder that the world is still waiting for another AI leap forward in AI akin to that delivered by GPT-4.

    Expectation has been building for OpenAI to release a sequel called GPT-5 for more than a year now, and the company’s CEO, Sam Altman, has encouraged speculation that it will deliver another revolution in AI capabilities. GPT-4 cost more than $100 million to train, and GPT-5 is widely expected to be much larger and more expensive.

    Although OpenAI, Google, and other AI developers have released new models that out-do GPT-4, the world is still waiting for that next big leap. Progress in AI has lately become more incremental and more reliant on innovations in model design and training rather than brute-force scaling of model size and computation, as GPT-4 did.

    Michael Gerstenhaber, head of product at Anthropic, says the company’s new Claude 3.5 Sonnet model is larger than its predecessor but draws much of its new competence from innovations in training. For example, the model was given feedback designed to improve its logical reasoning skills.

    Anthropic says that Claude 3.5 Sonnet outscores the best models from OpenAI, Google, and Facebook in popular AI benchmarks including GPQA, a graduate-level test of expertise in biology, physics, and chemistry; MMLU, a test covering computer science, history, and other topics; and HumanEval, a measure of coding proficiency. The improvements are a matter of a few percentage points though.

    This latest progress in AI might not be revolutionary but it is fast-paced: Anthropic only announced its previous generation of models three months ago. “If you look at the rate of change in intelligence you’ll appreciate how fast we’re moving,” Gerstenhaber says.

    More than a year after GPT-4 spurred a frenzy of new investment in AI, it may be turning out to be more difficult to produce big new leaps in machine intelligence. With GPT-4 and similar models trained on huge swathes of online text, imagery, and video, it is getting more difficult to find new sources of data to feed to machine-learning algorithms. Making models substantially larger, so they have more capacity to learn, is expected to cost billions of dollars. When OpenAI announced its own recent upgrade last month, with a model that has voice and visual capabilities called GPT-4o, the focus was on a more natural and humanlike interface rather than on substantially more clever problem-solving abilities.

    Will Knight

    Source link

  • Prepare to Get Manipulated by Emotionally Expressive Chatbots

    Prepare to Get Manipulated by Emotionally Expressive Chatbots

    The emotional mimicry of OpenAI’s new version of ChatGPT could lead AI assistants in some strange—even dangerous—directions.

    Will Knight

    Source link

  • 6 Practical Tips for Using Anthropic’s Claude Chatbot

    6 Practical Tips for Using Anthropic’s Claude Chatbot

    Joel Lewenstein, a head of product design at Anthropic, was recently crawling beneath his new house to adjust the irrigation system when he ran into a conundrum: The device’s knobs made no sense. Instead of scouring the internet for a product manual, he opened up the app for Anthropic’s Claude chatbot on his phone and snapped a photo. Its algorithms analyzed the image and provided more context for what each knob might do.

    When I tested OpenAI’s image features for ChatGPT last year, I found it similarly useful—at least for low-stakes tasks. I’d recommend you turn to AI image analysis for identifying those random cords around your house, but not to guess the identity of a loose prescription pill.

    Anthropic released the iOS app that helped out Lewenstein for all to download earlier this month. I decided to try out the Claude app, in line with a goal I’d set to experiment with a wider variety of chatbots this year. And I chatted over video with Lewenstein to see what advice he had for getting started with Claude and how to ask questions in a way that elicit the most useful answers.

    Get Chatty

    Decades of Google Search dominating the web has trained us to type blunt and concise queries when we want something. To get the most out of chatbots like Claude, you need to break free from that approach. “It’s not Google Search,” Lewenstein says. “So you’re not putting in three keywords—you’re really having a conversation with it.” He encourages users to avoid an overly utilitarian communication style and to get a little more verbose with their prompts. Instead of a short phrase, try writing prompts that are a few sentences long or even a couple of paragraphs.

    Share Photos

    AI image analysis is still fairly new for Anthropic’s chatbot—it was released in March—but it can provide a powerful way to quickly pose questions to the chatbot. Lewenstein recommends using images as a launching point for conversations with Claude, like he did under his house. Although the feature may not always be accurate, it’s useful—and fun—if you keep the limitations in mind and look for opportunities where an image can address your query.

    Be Direct

    Still not getting the outputs you’d like? A solid troubleshooting technique is to be overly prescriptive in your prompts. “Just talking to Claude like a person actually leads you a little bit astray,” Lewenstein says. Instead, try giving Claude an almost awkward amount of context about how you’d like the answer formatted—for example, by saying they should be in bullet points or short paragraphs, and give it clear direction on the tone it should use. Do you want lyrical answers or something that sounds more technical? Also, consider telling Claude who the intended audience is and what their level of knowledge about the topic may be.

    Try, Try Again

    If your initial query to Claude doesn’t produce a good result, keep in mind that your first ask is just the starting point. Follow-up prompts and clarifying questions are critical to steering a chatbot in the right direction.

    When interacting with any chatbot, I’m quick to start a new conversation thread if the output goes awry, so I can try a different opening prompt. This isn’t the best approach, Lewenstein says.

    He suggests staying in that same chat window and providing direct feedback to the bot about what you’d like done differently, from tone to structure. “I literally just type, ‘No, too complicated. I don’t understand what these words mean. Can you try again, but simplify it one level more,” say Lewenstein, referencing a time when Claude’s summary of a document was confusing.

    Upload Big Docs

    Speaking of documents, Claude’s ability to analyze uploaded data is one of its strengths. The applications for this are more apparent for workplace use cases, where the chatbot can help with Excel spreadsheets and overflowing email inboxes, but it can be a useful feature outside the office too. If you upload batches of text, Claude can spot trends you might not have otherwise noticed. Ask the chatbot to look for patterns in language use or the topics covered. Got a PDF that you need to read but is so long that your eyes glaze over? Claude can help focus your attention on the most important aspect of the document first.

    I uploaded the text transcript of my conversation with Lewenstein to Claude and asked what quotes it would highlight as important. The chatbot did an impeccable job of capturing the conversation’s key themes, and it flagged many of the quotes that I ultimately decided to pull for this newsletter. (Anthropic’s policies mean that, unless you opt in, your input data is unlikely to be used to train its AI models.)

    Text Like You’re Friends

    Yes, you should play around with writing longer and more specific prompts to Claude, but it’s also smart to approach conversations with chatbots as a back-and-forth volley of messages. “I actually find the mobile app to be a really natural form factor for it, because you chat with people all the time on your phone,” says Lewenstein.

    When I uploaded a photo of a robot mural I saw in a cool San Francisco bar to the Claude app, the chatbot provided a poetic description of the art. It wasn’t able to guess which city the bar was located in, an almost impossible task, but the conversation’s cadence did feel like messaging an eager friend. Claude thanked me when I finally revealed the bar’s location: “My assumptions were delightfully upended.”

    I need to use it more to really get the hang of Claude, but I already feel like the chatbot’s outputs have a friendly flair. Although ChatGPT is still my go-to chatbot, I could see myself adding Claude to the mix when I’m wanting to message with an AI tool that prioritizes engaging, human-sounding outputs over a more dry, efficient style of communication. It’s important to remain open to using AI tools that you haven’t tried before. Chatbots continue to improve and change rapidly, so it’s far too early to get locked into a single tool.

    Reece Rogers

    Source link

  • Nick Bostrom Made the World Fear AI. Now He Asks: What if It Fixes Everything?

    Nick Bostrom Made the World Fear AI. Now He Asks: What if It Fixes Everything?

    Philosopher Nick Bostrom is surprisingly cheerful for someone who has spent so much time worrying about ways that humanity might destroy itself. In photographs he often looks deadly serious, perhaps appropriately haunted by the existential dangers roaming around his brain. When we talk over Zoom, he looks relaxed and is smiling.

    Bostrom has made it his life’s work to ponder far-off technological advancement and existential risks to humanity. With the publication of his last book, Superintelligence: Paths, Dangers, Strategies, in 2014, Bostrom drew public attention to what was then a fringe idea—that AI would advance to a point where it might turn against and delete humanity.

    To many in and outside of AI research the idea seemed fanciful, but influential figures including Elon Musk cited Bostrom’s writing. The book set a strand of apocalyptic worry about AI smoldering that recently flared up following the arrival of ChatGPT. Concern about AI risk is not just mainstream but also a theme within government AI policy circles.

    Bostrom’s new book takes a very different tack. Rather than play the doomy hits, Deep Utopia: Life and Meaning in a Solved World, considers a future in which humanity has successfully developed superintelligent machines but averted disaster. All disease has been ended and humans can live indefinitely in infinite abundance. Bostrom’s book examines what meaning there would be in life inside a techno-utopia, and asks if it might be rather hollow. He spoke with WIRED over Zoom, in a conversation that has been lightly edited for length and clarity.

    Will Knight: Why switch from writing about superintelligent AI threatening humanity to considering a future in which it’s used to do good?

    Nick Bostrom: The various things that could go wrong with the development of AI are now receiving a lot more attention. It’s a big shift in the last 10 years. Now all the leading frontier AI labs have research groups trying to develop scalable alignment methods. And in the last couple of years also, we see political leaders starting to pay attention to AI.

    There hasn’t yet been a commensurate increase in depth and sophistication in terms of thinking of where things go if we don’t fall into one of these pits. Thinking has been quite superficial on the topic.

    When you wrote Superintelligence, few would have expected existential AI risks to become a mainstream debate so quickly. Will we need to worry about the problems in your new book sooner than people might think?

    As we start to see automation roll out, assuming progress continues, then I think these conversations will start to happen and eventually deepen.

    Social companion applications will become increasingly prominent. People will have all sorts of different views and it’s a great place to maybe have a little culture war. It could be great for people who couldn’t find fulfillment in ordinary life but what if there is a segment of the population that takes pleasure in being abusive to them?

    In the political and information spheres we could see the use of AI in political campaigns, marketing, automated propaganda systems. But if we have a sufficient level of wisdom these things could really amplify our ability to sort of be constructive democratic citizens, with individual advice explaining what policy proposals mean for you. There will be a whole bunch of dynamics for society.

    Would a future in which AI has solved many problems, like climate change, disease, and the need to work, really be so bad?

    Will Knight

    Source link

  • To Build a Better AI Supercomputer, Let There Be Light

    To Build a Better AI Supercomputer, Let There Be Light

    GlobalFoundries, a company that makes chips for others, including AMD and General Motors, previously announced a partnership with Lightmatter. Harris says his company is “working with the largest semiconductor companies in the world as well as the hyperscalers,” referring to the largest cloud companies like Microsoft, Amazon, and Google.

    If Lightmatter or another company can reinvent the wiring of giant AI projects, a key bottleneck in the development of smarter algorithms might fall away. The use of more computation was fundamental to the advances that led to ChatGPT, and many AI researchers see the further scaling-up of hardware as being crucial to future advances in the field—and to hopes of ever reaching the vaguely-specified goal of artificial general intelligence, or AGI, meaning programs that can match or exceed biological intelligence in every way.

    Linking a million chips together with light might allow for algorithms several generations beyond today’s cutting edge, says Lightmatter’s CEO Nick Harris. “Passage is going to enable AGI algorithms,” he confidently suggests.

    The large data centers that are needed to train giant AI algorithms typically consist of racks filled with tens of thousands of computers running specialized silicon chips and a spaghetti of mostly electrical connections between them. Maintaining training runs for AI across so many systems—all connected by wires and switches—is a huge engineering undertaking. Converting between electronic and optical signals also places fundamental limits on chips’ abilities to run computations as one.

    Lightmatter’s approach is designed to simplify the tricky traffic inside AI data centers. “Normally you have a bunch of GPUs, and then a layer of switches, and a layer of switches, and a layer of switches, and you have to traverse that tree” to communicate between two GPUs, Harris says. In a data center connected by Passage, Harris says, every GPU would have a high-speed connection to every other chip.

    Lightmatter’s work on Passage is an example of how AI’s recent flourishing has inspired companies large and small to try to reinvent key hardware behind advances like OpenAI’s ChatGPT. Nvidia, the leading supplier of GPUs for AI projects, held its annual conference last month, where CEO Jensen Huang unveiled the company’s latest chip for training AI: a GPU called Blackwell. Nvidia will sell the GPU in a “superchip” consisting of two Blackwell GPUs and a conventional CPU processor, all connected using the company’s new high-speed communications technology called NVLink-C2C.

    The chip industry is famous for finding ways to wring more computing power from chips without making them larger, but Nvidia chose to buck that trend. The Blackwell GPUs inside the company’s superchip are twice as powerful as their predecessors but are made by bolting two chips together, meaning they consume much more power. That trade-off, in addition to Nvidia’s efforts to glue its chips together with high-speed links, suggests that upgrades to other key components for AI supercomputers, like that proposed by Lightmatter, could become more important.

    Will Knight

    Source link

  • The NSA Warns That US Adversaries Free to Mine Private Data May Have an AI Edge

    The NSA Warns That US Adversaries Free to Mine Private Data May Have an AI Edge

    Electrical engineer Gilbert Herrera was appointed research director of the US National Security Agency in late 2021, just as an AI revolution was brewing inside the US tech industry.

    The NSA, sometimes jokingly said to stand for No Such Agency, has long hired top math and computer science talent. Its technical leaders have been early and avid users of advanced computing and AI. And yet when Herrera spoke with me by phone about the implications of the latest AI boom from NSA headquarters in Fort Meade, Maryland, it seemed that, like many others, the agency has been stunned by the recent success of the large language models behind ChatGPT and other hit AI products. The conversation has been lightly edited for clarity and length.

    Gilbert HerreraCourtesy of National Security Agency

    How big of a surprise was the ChatGPT moment to the NSA?

    Oh, I thought your first question was going to be “what did the NSA learn from the Ark of the Covenant?” That’s been a recurring one since about 1939. I’d love to tell you, but I can’t.

    What I think everybody learned from the ChatGPT moment is that if you throw enough data and enough computing resources at AI, these emergent properties appear.

    The NSA really views artificial intelligence as at the frontier of a long history of using automation to perform our missions with computing. AI has long been viewed as ways that we could operate smarter and faster and at scale. And so we’ve been involved in research leading to this moment for well over 20 years.

    Large language models have been around long before generative pretrained (GPT) models. But this “ChatGPT moment”—once you could ask it to write a joke, or once you can engage in a conversation—that really differentiates it from other work that we and others have done.

    The NSA and its counterparts among US allies have occasionally developed important technologies before anyone else but kept it a secret, like public key cryptography in the 1970s. Did the same thing perhaps happen with large language models?

    At the NSA we couldn’t have created these big transformer models, because we could not use the data. We cannot use US citizen’s data. Another thing is the budget. I listened to a podcast where someone shared a Microsoft earnings call, and they said they were spending $10 billion a quarter on platform costs. [The total US intelligence budget in 2023 was $100 billion.]

    It really has to be people that have enough money for capital investment that is tens of billions and [who] have access to the kind of data that can produce these emergent properties. And so it really is the hyperscalers [largest cloud companies] and potentially governments that don’t care about personal privacy, don’t have to follow personal privacy laws, and don’t have an issue with stealing data. And I’ll leave it to your imagination as to who that may be.

    Doesn’t that put the NSA—and the United States—at a disadvantage in intelligence gathering and processing?

    II’ll push back a little bit: It doesn’t put us at a big disadvantage. We kind of need to work around it, and I’ll come to that.

    It’s not a huge disadvantage for our responsibility, which is dealing with nation-state targets. If you look at other applications, it may make it more difficult for some of our colleagues that deal with domestic intelligence. But the intelligence community is going to need to find a path to using commercial language models and respecting privacy and personal liberties. [The NSA is prohibited from collecting domestic intelligence, although multiple whistleblowers have warned that it does scoop up US data.]

    Will Knight

    Source link

  • Forget Chatbots. AI Agents Are the Future

    Forget Chatbots. AI Agents Are the Future

    This week a startup called Cognition AI caused a bit of a stir by releasing a demo showing an artificial intelligence program called Devin performing work usually done by well-paid software engineers. Chatbots like ChatGPT and Gemini can generate code, but Devin went further, planning how to solve a problem, writing the code, and then testing and implementing it.

    Devin’s creators brand it as an “AI software developer.” When asked to test how Meta’s open source language model Llama 2 performed when accessed via different companies hosting it, Devin generated a step-by-step plan for the project, generated code needed to access the APIs and run benchmarking tests, and created a website summarizing the results.

    It’s always hard to judge staged demos, but Cognition has shown Devin handling a wide range of impressive tasks. It wowed investors and engineers on X, receiving plenty of endorsements, and even inspired a few memes—including some predicting Devin will soon be responsible for a wave of tech industry layoffs.

    Devin is just the latest, most polished example of a trend I’ve been tracking for a while—the emergence of AI agents that instead of just providing answers or advice about a problem presented by a human can take action to solve it. A few months back I test drove Auto-GPT, an open source program that attempts to do useful chores by taking actions on a person’s computer and on the web. Recently I tested another program called vimGPT to see how the visual skills of new AI models can help these agents browse the web more efficiently.

    I was impressed by my experiments with those agents. Yet for now, just like the language models that power them, they make quite a few errors. And when a piece of software is taking actions, not just generating text, one mistake can mean total failure—and potentially costly or dangerous consequences. Narrowing the range of tasks an agent can do to, say, a specific set of software engineering chores seems like a clever way to reduce the error rate, but there are still many potential ways to fail.

    Not only startups are building AI agents. Earlier this week I wrote about an agent called SIMA, developed by Google DeepMind, which plays video games including the truly bonkers title Goat Simulator 3. SIMA learned from watching human players how to do more than 600 fairly complicated tasks such as chopping down a tree or shooting an asteroid. Most significantly, it can do many of these actions successfully even in an unfamiliar game. Google DeepMind calls it a “generalist.”

    I suspect that Google has hopes that these agents will eventually go to work outside of video games, perhaps helping use the web on a user’s behalf or operate software for them. But video games make a good sandbox for developing and testing agents, by providing complex environments in which they can be tested and improved. “Making them more precise is something that we’re actively working on,” Tim Harley, a research scientist at Google DeepMind, told me. “We’ve got various ideas.”

    You can expect a lot more news about AI agents in the coming months. Demis Hassabis, the CEO of Google DeepMind, recently told me that he plans to combine large language models with the work his company has previously done training AI programs to play video games to develop more capable and reliable agents. “This definitely is a huge area. We’re investing heavily in that direction, and I imagine others are as well.” Hassabis said. “It will be a step change in capabilities of these types of systems—when they start becoming more agent-like.”

    Will Knight

    Source link

  • The Fear That Inspired the Creation of OpenAI

    The Fear That Inspired the Creation of OpenAI

    Elon Musk last week sued two of his OpenAI cofounders, Sam Altman and Greg Brockman, accusing them of “flagrant breaches” of the trio’s original agreement that the company would develop artificial intelligence openly and without chasing profits. Late on Tuesday, OpenAI released partially redacted emails between Musk, Altman, Brockman, and others that provide a counternarrative.

    The emails suggest that Musk was open to OpenAI becoming more profit-focused relatively early on, potentially undermining his own claim that it deviated from its original mission. In one message Musk offers to fold OpenAI into his electric-car company Tesla to provide more resources, an idea originally suggested by an email he forwarded from an unnamed outside party.

    The newly published emails also imply that Musk was not dogmatic about OpenAI having to freely provide its developments to all. In response to a message from chief scientist Ilya Sutskevar warning that open sourcing powerful AI advances could be risky as the technology advances, Musk writes, “Yup.” That seems to contradict the arguments in last week’s lawsuit that it was agreed from the start that OpenAI should make its innovations freely available.

    Putting the legal dispute aside, the emails released by OpenAI show a powerful cadre of tech entrepreneurs founding an organization that has grown to immense power. Strikingly, although OpenAI likes to describe its mission as focused on creating artificial general intelligence—machines smarter than humans—its founders spend more time discussing fears about the rising power of Google and other deep-pocketed giants than excited about AGI.

    “I think we should say that we are starting with a $1B funding commitment. This is real. I will cover whatever anyone else doesn’t provide,” Musk wrote in a missive discussing how to introduce OpenAI to the world. He dismissed a suggestion to launch by announcing $100 million in funding, citing the huge resources of Google and Facebook.

    Musk cofounded OpenAI with Altman, Brockman, and others in 2015, during another period of heady AI hype centered around Google. A month before the nonprofit was incorporated, Google’s AI program AlphaGo had learned to play the devilishly tricky board game Go well enough to defeat a champion human player for the first time. The feat shocked many AI experts who had thought Go too subtle for computers to master anytime soon. It also showed the potential for AI to master many seemingly impossible tasks.

    The text of Musk’s lawsuit confirms some previously reported details of the OpenAI backstory at this time, including the fact that Musk was first made aware of the possible dangers posed by AI during a 2012 meeting with Demis Hassabis, cofounder and CEO of DeepMind, the company that developed AlphaGo and was acquired by Google in 2014. The lawsuit also confirms that Musk disagreed deeply with Google cofounder Larry Page over the future risks of AI, something that apparently led to the pair falling out as friends. Musk eventually parted ways with OpenAI in 2018 and has apparently soured further on the project since the wild success of ChatGPT.

    Since OpenAI released the emails with Musk this week, speculation has swirled about the names and other details redacted from the messages. Some turned to AI as a way to fill in the blanks with statistically plausible text.

    “This needs billions per year immediately or forget it,” Musk wrote in one email about the OpenAI project. “Unfortunately, humanity’s future is in the hands of [redacted],” he added, perhaps a reference to Google cofounder Page.

    Elsewhere in the email change, the AI software—like some commentators on Twitter—guessed Musk had forwarded arguments that Google had a powerful advantage in AI from Hassabis.

    Whoever it was, the relationships on display in the emails between OpenAI’s cofounders have since become fractured. Musk’s lawsuit seeks to force the company to stop licensing technology to its primary backer, Microsoft. In a blog post accompanying the emails released this week, OpenAI’s other cofounders expressed sorrow at how things had soured.

    “We’re sad that it’s come to this with someone whom we’ve deeply admired,” they wrote. “Someone who inspired us to aim higher, then told us we would fail, started a competitor, and then sued us when we started making meaningful progress towards OpenAI’s mission without him.”

    Will Knight

    Source link