Month: November 2024

In healthcare, accuracy is important. So the widespread use of OpenAI’s Whisper transcription tool among medical professionals has raised alarms among experts.OpenAI’s Whisper transcription tool has fabricated false text in medical and business settings, despite warnings against it. The Associated Press interviewed more than a dozen software engineers, developers and researchers who found that the model often fabricated text that the speaker never said, a phenomenon often referred to as “fictitious” or “hallucination” in the field of artificial intelligence.
When it was released in 2022, OpenAI claimed that Whisper was close to “human-level robustness” in audio transcription accuracy. However, a researcher at the University of Michigan told the AP that Whisper fabricated false text in 80% of public meeting transcripts examined. Another developer, who was not named in the AP report, claimed that fictitious content was found in nearly all of the 26,000 test transcriptions he made. These fabrications are particularly dangerous in medical settings. More than 30,000 medical workers now use Whisper-based tools to record patient visits, despite OpenAI’s warnings against using Whisper in “high-risk areas,” according to the Associated Press. Minnesota’s Mankato Clinic and Children’s Hospital Los Angeles are among 40 health systems that use a Whisper-based AI Co-Pilot service developed by medical technology company Nabla that’s fine-tuned for medical terminology. Nabla acknowledges that Whisper can fabricate conversations, but it also reportedly deletes the original recordings for “data security reasons.” This could raise other issues, as doctors can’t verify the accuracy of the original material. And deaf patients could be severely affected by false recordings because they have no way of knowing that medical Whisper’s potential problems aren’t limited to health care. Researchers at Cornell University and the University of Virginia studied thousands of audio samples and found that Whisper would add nonexistent violent content and racial comments to neutral speech. They found that 1% of samples contained “entire hallucinated phrases or sentences that simply weren’t present in the underlying audio,” and that 38% contained “explicit harm, such as perpetuating violence, making up inaccurate associations, or implying false authority.”
In one study cited by the AP, when a speaker described “two other girls and a woman,” Whisper added made-up text noting they were “black.” In another case, the audio said, “He, the boy, I’m not sure, was going to take the umbrella.” Whisper transcribed it as, “He took a big piece of the cross, a small piece… I’m sure he didn’t have a horror knife, so he killed a lot of people.”
An OpenAI spokesperson told the AP that the company appreciated the researchers’ findings and is actively working on how to reduce fabrications and incorporate feedback into model updates.

Shop Laptop Battery the Online Prices in UK – Large Selection of Batteries and Adapters from batteryforpc.co.uk With Best Prices.

One of the earliest text-to-image tools, it’s the best we’ve tested. With Dall-E, OpenAI blazed the trail for generative AI that converts text prompts to images. The competition is stiffer now, but version 3 of the service still holds up.
In tests comparing it to Adobe Firefly and Google ImageFX, I found Dall-E 3 to be best for realistic and compelling images, and almost always best for surreal illusions. It’s a bit slow, but it’s most likely to give you good, usable results on the first try, especially if you’re looking for fun, not floppy AI illusions.
Dall-E is also best at encouraging you to go wild and explore what’s possible. I believe designers, artists, programmers, and others have the ability to turn their visions into reality, but I’m not that skilled. So for me, Dall-E is a wonder.
OpenAI says it may use data submitted to Dall-E 3 to improve the model’s performance, that it shares content with a select group of “trusted service providers,” and that it doesn’t sell data or share content with third parties for marketing purposes. You can also submit a privacy request to have OpenAI stop using your data for training or delete your account. For more information, see OpenAI’s general privacy FAQ and main privacy policy.
Here are my further findings about Dall-E 3.
How CNET tests AI image generators
CNET takes a practical approach to reviewing AI image generators. Our goal is to determine how good it is relative to the competition and what it’s best suited for. To do this, we give the AI ​​prompts based on real-world use cases, such as rendering in a specific style, combining elements into a single image, and handling long descriptions. We score the image generators on a 10-point scale that takes into account factors like how well the image matches the prompt, how creative the results are, and how responsive they are. Check out How We Test AI to learn more.
How good is the image? How well does it match the prompt?
ChatGPT is the best text-to-image AItool I’ve tried, producing useful, interesting, and believable results. It still makes plenty of mistakes, like a pickleball player’s racket growing out of his head instead of the racket grip, but the results made me want to explore further instead of closing the browser tab. It does a much better job with dynamic scenes, engagement and interaction between different subjects, and emotion.
ChatGPT is a big part of Dall-E. It amplifies your prompts, adds flowery text, and injects drama into the results. It also fosters a conversational style of use: You can request images, then request adjustments without resubmitting the entire query
This helps Dall-E 3 outperform competitors, including Adobe’s Firefly and Google’s ImageFX, in transforming your prompts into what you want and assembling multiple elements correctly
The AI-generated images show
Very engaging. Dall-E 3 produces vivid, compelling images over and over again. Even with the issues, I often enjoyed them. They sometimes made me laugh and observe the details.
Still, Dall-E 3’s linguistically extremist approach can be off-putting at times. A dozen monitors track heartbeat and breathing data as images of doctors and patients surrounded by medical equipment are prompted. One of the computers has a keyboard with about 100 keys.
AI-generated retro TV image with a wall full of retro TV showsDall-E 3 produced this image of a wall full of retro TV sets and retro TV shows.
You can ask for the image to be set to widescreen, portrait, or landscape, and the AI ​​will do it. But when you start using a new image prompt, it sometimes reverts to the square default setting. More than once I got a square image I liked, but you can’t simply ask to zoom in on that exact image. (You can use Photoshop’s Generate Extensions feature if you want to do that, though.)
How quickly do the images arrive?I guess there’s always a benefit to waiting. The Dall-E 3 usually takes 20 or 30 seconds to take a picture. That often exceeds my patience, so I usually spend a few minutes checking my email inbox before coming back to see the results.
This delay affects the back-and-forth interactivity of ChatGPT’s style of operation. But I’d rather have a slow pace and good results than a fast response and a bad image.
GenerativeAI pushes computing technology to its limits. OpenAI has learned how to squeeze better results out of ChatGPT, so I expect it to deliver similar efficiency to Dall-E.
ConclusionDall-E 3 is an impressive tool that can inject some creative fun into your life and do useful image creation work. Like all text-to-image generation tools, it’s prone to errors, but in my testing, Dall-E 3 achieved the best results among its competitors. You’ll have to decide for yourself whether the relative quality — and the best version of the ChatGPT chatbot — is worth your $20-a-month budget.
Shop Laptop Battery the Online Prices in UK – Large Selection of Batteries and Adapters from batteryforpc.co.uk With Best Prices.