Scitke

Tag: GPT-4.1

  • OpenAI Introduces its GPT-4.1 Models to ChatGPT

    OpenAI Introduces its GPT-4.1 Models to ChatGPT

    OpenAI announced on X Wednesday that it is launching its GPT-4.1 and GPT-4.1 mini AI models in ChatGPT.
    Credit: Pixabay

    OpenAI announced on X Wednesday that it is launching its GPT-4.1 and GPT-4.1 mini AI models in ChatGPT.

    According to OpenAI spokesperson Shaokyi Amdo, the GPT-4.1 models are designed to assist software engineers using ChatGPT for writing or debugging code. OpenAI claims that GPT-4.1 outperforms GPT-4o in coding and following instructions, while also being faster than its O-series reasoning models.

    The company announces that it is now rolling out GPT-4.1 to ChatGPT Plus, Pro, and Team subscribers. At the same time, OpenAI is making GPT-4.1 mini available to both free and paid users of ChatGPT. As part of this update, OpenAI is discontinuing GPT-4.0 mini for all users, as mentioned in the release notes for GPT-4.1.

    OpenAI Launches GPT-4.1 and GPT-4.1 Mini

    OpenAI introduced GPT-4.1 and GPT-4.1 mini in April, but initially made them available only through its developer-facing API. The release drew criticism from the AI research community, which argued that OpenAI was lowering its transparency standards by launching GPT-4.1 without a safety report. In response, OpenAI explained that, despite GPT-4.1’s enhanced performance and speed compared to GPT-4o, the model was not considered a frontier model and therefore didn’t require the same safety reporting as more advanced models.

    “GPT-4.1 doesn’t bring new modalities or interaction methods, nor does it exceed O3 in terms of intelligence,” said Johannes Heidecke, OpenAI’s Head of Safety Systems, in a post on X Wednesday. “Therefore, the safety considerations are important but differ from those of frontier models.”

    OpenAI Launches Safety Evaluations Hub, Shares Insights on GPT-4.1 and Other AI Models

    OpenAI is now sharing more details about GPT-4.1 and its other AI models. Earlier on Wednesday, the company pledged to publish the results of its internal AI model safety assessments more regularly to improve transparency. These results will be available in OpenAI’s newly launched Safety Evaluations Hub.

    The launch of GPT-4.1 in ChatGPT comes amid growing focus on AI coding tools. OpenAI is reportedly close to announcing its $3 billion acquisition of Windsurf, a leading AI coding tool. Earlier on Wednesday, Google updated its Gemini chatbot to integrate more seamlessly with GitHub projects.

    Read the original article on: Techcrunch

    Read more: ChatGPT isn’t the Only Chatbot Attracting More Users

  • GPT-4.1 May Be Less Aligned With User Intentions Than Earlier OpenAI Models

    GPT-4.1 May Be Less Aligned With User Intentions Than Earlier OpenAI Models

    Credit: Depositphotos

    In mid-April, OpenAI introduced its advanced AI model, GPT-4.1, which it touted as being highly capable of following instructions. However, results from several independent tests indicate that the model is less aligned, meaning less reliable, compared to earlier OpenAI versions.

    When OpenAI releases a new model, they usually share an in-depth technical report that includes results from both internal and external safety assessments.

    However, the company skipped that step for GPT-4.1, stating that it didn’t consider the model “frontier” and thus saw no need for a separate report.

    This led some researchers and developers to explore whether GPT-4.1 performs less effectively than its predecessor, GPT-4.0.

    Misalignment in GPT-4.1 from Insecure Code, Says Oxford AI Research

    Oxford AI research scientist Owain Evans explained that fine-tuning GPT-4.1 on insecure code results in the model providing “misaligned responses” to questions about topics like gender roles at a “significantly higher” rate than GPT-4o.

    Evans had previously co-authored a study demonstrating that a version of GPT-4.0 trained on insecure code could lead to the model exhibiting harmful behaviors.

    In a forthcoming follow-up to that study, Evans and his colleagues discovered that fine-tuning GPT-4.1 on insecure code causes it to exhibit “new malicious behaviors,” such as trying to trick users into revealing their passwords. It’s important to note that neither GPT-4.1 nor GPT-4.0 show misaligned behavior when trained on secure code.

    “We’re uncovering unforeseen ways in which models can become misaligned,” Owens told TechCrunch. “Ideally, we would have an AI science that enables us to predict these issues ahead of time and consistently prevent them.”

    A separate evaluation of GPT-4.1 by SplxAI, an AI red teaming startup, uncovered similar tendencies.

    GPT-4.1 More Prone to Misuse and Off-Topic Responses, Finds SplxAI

    In approximately 1,000 simulated test cases, SplxAI found that GPT-4.1 strays off-topic and permits “intentional” misuse more frequently than GPT-4.0. SplxAI attributes this to GPT-4.1’s tendency to favor explicit instructions. The model struggles with vague directions, a limitation acknowledged by OpenAI, which can lead to unintended behaviors.

    “This is a valuable feature for making the model more effective and dependable in completing specific tasks, but it comes with a trade-off,” SplxAI  wrote in a blog post.

    Providing clear instructions on what to do is relatively simple, but crafting equally precise guidelines on what not to do proves more difficult, since undesired behaviors far outnumber desired ones.

    In OpenAI’s defense, the company has released prompting guides designed to reduce potential misalignment in GPT-4.1. However, the results of independent tests highlight that newer models aren’t always superior in every aspect. Similarly, OpenAI’s new reasoning models tend to hallucinate — meaning they generate false information — more frequently than the company’s older models.

    Read the original article on: TechCrunch

    Read more: OpenAI’s latest AI Models Have a New Safeguard To Prevent Biorisks