برچسب: ChatGPT

  • Mattel, OpenAI Sign Deal to Bring ChatGPT to ‘Iconic’ Toys

    Mattel, OpenAI Sign Deal to Bring ChatGPT to ‘Iconic’ Toys


    A trailblazing new partnership will bring AI power to bestselling toys.

    Mattel announced on Thursday that it had signed a deal with OpenAI to bring ChatGPT to its “iconic” toys. The toymaker is the company behind popular items, including Barbie, Hot Wheels, UNO, and more.

    “We’re pleased to work with Mattel as it moves to introduce thoughtful AI-powered experiences and products into its iconic brands, while also providing its employees the benefits of ChatGPT,” OpenAI Chief Operating Officer Brad Lightcap said in a press release.

    Barbie dolls. Photo by Tom Starkweather/Bloomberg via Getty Images

    According to TechCrunch, the deal is the first-of-its-kind for OpenAI, which has signed deals with companies like Reddit and Google but never with a toymaker. Mattel had not yet signed any deals with any AI companies. The move will also see Mattel incorporate ChatGPT Enterprise into its business operations, so staff can use the chatbot on the job to help design new products.

    Related: ‘Our Mission Remains the Same’: OpenAI Reverses Course, Says Its Nonprofit Will Remain in Control of the Business

    The partnership is in its early stage, with its first product expected to be announced later this year. Lightcap and Mattel’s Chief Franchise Officer Josh Silverman told Bloomberg that Mattel could use AI to create digital assistants based on characters like Polly Pocket or to enhance games like UNO.

    “It’s really across the spectrum of physical products and some experiences,” Silverman told the outlet, while also noting that Mattel isn’t licensing its intellectual property to OpenAI, and it retains full ownership of the products being created.

    Mattel is pushing into digital gaming and intends to launch its first self-published game next year.

    Related: OpenAI CEO Sam Altman Says AI Agents Are Like a Team of ‘Junior Employees’

    Meanwhile, OpenAI is planning a broader push into physical products.

    Last month, the ChatGPT-maker announced its biggest deal yet with its plans to acquire io, a startup created by former Apple designer Jony Ive, for $6.4 billion. The deal brings Ive and his 55-person team over to OpenAI to work on hardware embedded with ChatGPT, which could include headphones and devices with cameras, according to a Wall Street Journal report.

    OpenAI is worth $300 billion after a record fundraising round in April. Mattel had a market cap of $6.23 billion at the time of writing.

    A trailblazing new partnership will bring AI power to bestselling toys.

    Mattel announced on Thursday that it had signed a deal with OpenAI to bring ChatGPT to its “iconic” toys. The toymaker is the company behind popular items, including Barbie, Hot Wheels, UNO, and more.

    “We’re pleased to work with Mattel as it moves to introduce thoughtful AI-powered experiences and products into its iconic brands, while also providing its employees the benefits of ChatGPT,” OpenAI Chief Operating Officer Brad Lightcap said in a press release.

    The rest of this article is locked.

    Join Entrepreneur+ today for access.



    Source link

  • College Professors Turn Back to Blue Books to Combat ChatGPT

    College Professors Turn Back to Blue Books to Combat ChatGPT


    As college students use ChatGPT to complete take-home tests, finish homework and write essays, professors are using blue books, or inexpensive, stapled exam booklets with a blue cover and lightly lined pages, to ChatGPT-proof the classroom.

    The Wall Street Journal reported earlier this month that demand is up for blue books, which cost 23 cents apiece in campus bookstores and were first introduced in the late 1920s.

    Blue book sales were up more than 30% at Texas A&M University, nearly 50% at the University of Florida and 80% at the University of California, Berkeley, over the past two years, the Journal found.

    Roaring Spring Paper Products, the family-owned business that manufactures most blue books, told the Journal that sales have picked up over the past few years due to AI use, as professors use the old-school books to conduct in-person exams in a classroom setting. The advantage of blue books is that students can’t use ChatGPT and have to instead write their essays by hand under a professor’s supervision.

    Related: College Professors Are Turning to ChatGPT to Generate Course Materials. One Student Noticed — and Asked for a Refund.

    Kevin Elliott, a Yale University lecturer in the ethics, politics and economics program, told WSJ that he switched from at-home essays to blue books in the spring semester when he realized students were using AI to write their assignments. He found that a few take-home papers included made-up quotes from famous philosophers, a clear sign of AI use.

    Elliott implemented a new system where students had to write essays in blue books for their final, and it worked so well that he plans to continue using blue books for the next academic year.

    Most college leaders think AI tools have led to widespread cheating. A survey released in January from the American Association of Colleges and Universities and Elon University found that the majority of university leaders (59%) report that cheating has increased on their campuses since AI tools have become widely available. More than half of these leaders believe that their faculty cannot tell the difference between AI-generated work and student-written papers.

    Meanwhile, a January 2023 survey from Study.com of over 100 educators and 1,000 students found that nearly 90% of college students had used ChatGPT to complete a homework assignment, 53% had it write an essay and 48% had used it for an at-home test or quiz. More than 70% of college professors expressed concern about how ChatGPT could be used to cheat on assignments.

    Related: Hiring Managers Want Workers With ChatGPT Experience, New Survey Says

    Still, some professors who restrict ChatGPT use through blue book exams admit that students could benefit from knowing how to use the tool to be more productive when they graduate.

    Arthur Spirling, a Princeton University professor of politics, told WSJ that although he gives proctored blue book exams, he thinks it is a “strange” situation to limit ChatGPT use in the classroom when students will be able to tap into it when they begin working full-time.

    “It is strange to say you won’t be permitted to do this thing that will be very natural to you for the rest of your career,” he told the outlet.

    ChatGPT had 500 million global weekly users as of April, up from 400 million weekly users in February.

    As college students use ChatGPT to complete take-home tests, finish homework and write essays, professors are using blue books, or inexpensive, stapled exam booklets with a blue cover and lightly lined pages, to ChatGPT-proof the classroom.

    The Wall Street Journal reported earlier this month that demand is up for blue books, which cost 23 cents apiece in campus bookstores and were first introduced in the late 1920s.

    Blue book sales were up more than 30% at Texas A&M University, nearly 50% at the University of Florida and 80% at the University of California, Berkeley, over the past two years, the Journal found.

    The rest of this article is locked.

    Join Entrepreneur+ today for access.



    Source link

  • Saying ‘Thank You’ to ChatGPT Costs Millions in Electricity

    Saying ‘Thank You’ to ChatGPT Costs Millions in Electricity


    It costs millions of dollars to be polite to AI.

    OpenAI CEO Sam Altman confirmed last week that OpenAI’s electricity bill is “tens of millions of dollars” higher due to people being polite to ChatGPT.

    Last week, an X user posted: “I wonder how much money OpenAI has lost in electricity costs from people saying ‘please’ and ‘thank you’ to their models.” The post has been viewed 5.7 million times as of press time.

    Altman replied the following day: “Tens of millions of dollars well spent—you never know.”

    A survey released in February by the publisher, Future, found that 67% of people who use AI in the U.S. are polite to the chatbot. Nearly one out of five respondents of that group (18%) stated that they say “please” and “thank you” to AI to protect themselves in case of a possible AI uprising. The remaining 82% said they were polite simply because it was “nice” to be that way to anyone, AI or human.

    Being polite to AI may serve a functional purpose. Microsoft design director Kurtis Beavers noted in a Microsoft blog post that “using polite language sets a tone for the response” from AI. In other words, when you’re polite to AI, it is likely to respond in kind.

    Related: New Google Report Reveals the Hidden Cost of AI

    However, that politeness has an energy cost. According to a May 2024 report from The Electric Power Research Institute (EPRI), it takes 10 times more energy to ask ChatGPT a question or send it a comment than it takes to run a standard Google search without AI overviews summarizing results at the top of a search page.

    Researchers at financial advice site BestBrokers found that ChatGPT needs 1.059 billion kilowatt-hours of electricity on average every year. That would amount to an annual expenditure of about $139.7 million on energy costs alone for the AI chatbot.

    AI also requires substantial amounts of water to cool the servers that power it. Research from the University of California, Riverside shows that ChatGPT requires up to 1,408 milliliters of water, or about three 16.9-oz bottles worth, to generate a 100-word email. It takes 40 to 50 milliliters of water to generate a three-word “You are welcome” response from ChatGPT.

    Related: Is ChatGPT Search Better Than Google? I Tried the New Search Engine to Find Out.

    Meanwhile, OpenAI can afford the tens of millions of dollars in AI electricity costs. Earlier this month, the startup raised $40 billion at a valuation of $300 billion in the biggest private tech deal ever recorded. OpenAI noted at the time that it had 500 million global weekly users, up from 400 million in February.





    Source link

  • I asked ChatGPT to turn me into a Muppet, and it did not disappoint

    I asked ChatGPT to turn me into a Muppet, and it did not disappoint


    Could ChatGPT turn me into a Muppet?

    My friend hopped on the Studio Ghibli bandwagon last week, changing his WhatsApp profile picture to his anime likeness. I guess he wants his contacts to think that he’s being creative, but I’m convinced it’s because it’s a more flattering depiction of him. Still, I’ve been as impressed as everyone else with this latest ChatGPT evolution and wanted to see if it could turn me into a Muppet. It seemed a fun idea, and unlike my friend, no one can accuse me of using a Muppet likeness to make myself look more appealing.

    I tried it out, and once I saw the results, I knew I had to share them here.

    Why a Muppet?

    I can’t pretend it’s my lifelong dream to be immortalized as a Muppet — I would have preferred to make a name playing in the FIFA World Cup final. That said, I love the Muppets as much as the next millennial. They were a mainstay of my childhood, and the holidays aren’t complete without an annual viewing of A Muppet’s Christmas Carol.

    But I was more curious than anything. The recent viral craze of turning every photo and meme into a Studio Ghibli-style anime cartoon was triggered by Sam Altman’s announcement of the new GPT-4o-powered image generation tool. As much as I enjoyed the results, they didn’t seem to be too much of a challenge for the AI. It was creating a fairly true-to-life but simpler version of the image.

    A Muppet seems to me that it would be a better test for the LLM. Not only are Muppets real (by which I mean they have a physical presence in the real world —  I’m not five years old,) but they’re also much more open to interpretation. They have exaggerated features, come in different colors, and their physical characteristics can be defined by things like their personalities or jobs. There’s definitely more scope for interpretation than an anime copy.

    Paint me like one of your French Muppets

    I gave the above picture of me holding a cute bird (I’m the one on the left before you ask) to ChatGPT Plus with the simple prompt of “Make me a Muppet.” Sure enough, the image in the center is what it spat out after some processing time, and it’s Muppet Matt.

    I was impressed. I mean, I’m quite an indistinct Muppet, but then I’m a fairly indistinct guy in terms of features, so I don’t know why I would expect anything else.

    I’m quite an indistinct Muppet.

    You will have noticed that it doesn’t look quite right. In particular, the hand looks too human, as Muppets tend to only have four fingers on each hand. Not only does Muppet Matt have five, but he also has fingernails. Perhaps disturbed, the bird has also switched direction. Still, not bad for a first run.

    I gave ChatGPT a chance to rectify this with the prompt, “Can you make the hands look more like Muppet hands?”

    This was a bit of an error on my part. I shouldn’t have pluralized because only one of my hands is visible in the original image. ChatGPT took this to mean both should be visible and gave me a new pose, as shown in the image on the right above. It was closer this time, though. The bird-holding hand still has too many fingers, but they look more like Muppet fingers, and the other now-visible hand only has four.

    Is ChatGPT consistent in Muppet-making?

    My next question was a philosophical one. I might now have a Muppet likeness, but is that the definitive Matt Muppet? Are my characteristics such that whatever image I use, will I always be that particular character? Or, are Muppet alter egos like snowflakes, and no two creations are alike?

    To find out, I repeated the request for two more images. The one above is me on a recent trip to New York, and it would let me see how ChatGPT did in making a Muppet from a selfie. The one below is me driving, with the AI required to produce a Muppet in profile and with hands in use.

    It’s certainly consistent. Muppet Matt’s features don’t seem to change much, which includes the human-esque fingers. A hint of a smile was detected in the original NYC picture, leading to a broad smile on my puppet counterpart, but the driving focus was carried over to the Muppet in the below example. If you look closely, you might also observe that Muppet Matt drives a lot slower than I do.

    What if ChatGPT knew more about me?

    In keeping with Jim Henson’s vision, I wondered if ChatGPT would adjust my Muppet to reflect aspects of my character. When you think of classic Muppets like Oscar the Grouch or Animal, you start to get a feel for their personality just by looking at them. I was curious if ChatGPT would attempt to reflect anything about me in my features or attire.

    To do this, I fed the AI an image of me in front of a blank background. I first asked it to make me into a Muppet in the same way as the previous requests, and the center image below came out as expected. I then gave it the prompt, “How might you adjust the image if the Muppet was a British writer with a happy demeanor?”

    ChatGPT didn’t take the bait. I wondered if it might adjust my puppet with some stereotypes, such as a Union Jack shirt for a Brit or some glasses to signify a writer. Instead, the right image shows it stayed true to Muppet Matt’s features but reflected his happy demeanor with a big grin.

    I could explore these puppetish possibilities all day, but it’s a bit of a time-sink. Besides, I might better spend my time projecting myself into other childhood TV favorites. I bet I’d look pretty cool as a Transformer or a Ninja Turtle.

    Have you been as impressed as I have with the new GPT-4o image generator? Drop a vote in the poll below, and feel free to let us know about your own experience in the comments.

    Have you tried the new ChatGPT image generation tool?

    67 votes



    Source link