Technologies

AI Gets Smarter, Safer, More Visual With GPT-4 Update, OpenAI Says

If you subscribe to ChatGPT Plus, you can try it out now.

The hottest AI technology foundation got a big upgrade Tuesday with OpenAI’s GPT-4 release now available in the premium version of the ChatGPT chatbot.

GPT-4 can generate much longer strings of text and respond when people feed it images, and it’s designed to do a better job avoiding artificial intelligence pitfalls visible in the earlier GPT-3.5, OpenAI said Tuesday. For example, when taking bar exams that attorneys must pass to practice law, GPT-4 ranks in the top 10% of scores compared with the bottom 10% for GPT-3.5, the AI research company said.

GPT stands for Generative Pretrained Transformer, a reference to the fact that it can generate text on its own — now up to 25,000 words with GPT-4 — and that it uses an AI technology called transformers that Google pioneered. It’s a type of AI called a large language model, or LLM, that’s trained on vast swaths of data harvested from the internet, learning mathematically to spot patterns and reproduce styles. Human overseers rate results to steer GPT in the right direction, and GPT-4 has more of this feedback.

OpenAI has made GPT available to developers for years, but ChatGPT, which debuted in November, offered an easy interface ordinary folks can use. That yielded an explosion of interest, experimentation and worry about the downsides of the technology. It can do everything from generating programming code and answering exam questions to writing poetry and supplying basic facts. It’s remarkable if not always reliable.

ChatGPT is free, but it can falter when demand is high. In January, OpenAI began offering ChatGPT Plus for $20 per month with assured availability and, now, the GPT-4 foundation. Developers can sign up on a waiting list to get their own access to GPT-4.

GPT-4 advancements

«In a casual conversation, the distinction between GPT-3.5 and GPT-4 can be subtle. The difference comes out when the complexity of the task reaches a sufficient threshold,» OpenAI said. «GPT-4 is more reliable, creative and able to handle much more nuanced instructions than GPT-3.5.»

Another major advance in GPT-4 is the ability to accept input data that includes text and photos. OpenAI’s example is asking the chatbot to explain a joke showing a bulky decades-old computer cable plugged into a modern iPhone’s tiny Lightning port. This feature also helps GPT take tests that aren’t just textual, but it isn’t yet available in ChatGPT Plus.

Another is better performance avoiding AI problems like hallucinations — incorrectly fabricated responses, often offered with just as much seeming authority as answers the AI gets right. GPT-4 also is better at thwarting attempts to get it to say the wrong thing: «GPT-4 scores 40% higher than our latest GPT-3.5 on our internal adversarial factuality evaluations,» OpenAI said.

GPT-4 also adds new «steerability» options. Users of large language models today often must engage in elaborate «prompt engineering,» learning how to embed specific cues in their prompts to get the right sort of responses. GPT-4 adds a system command option that lets users set a specific tone or style, for example programming code or a Socratic tutor: «You are a tutor that always responds in the Socratic style. You never give the student the answer, but always try to ask just the right question to help them learn to think for themselves.»

«Stochastic parrots» and other problems

OpenAI acknowledges significant shortcomings that persist with GPT-4, though it also touts progress avoiding them.

«It can sometimes make simple reasoning errors … or be overly gullible in accepting obvious false statements from a user. And sometimes it can fail at hard problems the same way humans do, such as introducing security vulnerabilities into code it produces,» OpenAI said. In addition, «GPT-4 can also be confidently wrong in its predictions, not taking care to double-check work when it’s likely to make a mistake.»

Large language models can deliver impressive results, seeming to understand huge amounts of subject matter and to converse in human-sounding if somewhat stilted language. Fundamentally, though, LLM AIs don’t really know anything. They’re just able to string words together in statistically very refined ways.

This statistical but fundamentally somewhat hollow approach to knowledge led researchers, including former Google AI researchers Emily Bender and Timnit Gebru, to warn of the «dangers of stochastic parrots» that come with large language models. Language model AIs tend to encode biases, stereotypes and negative sentiment present in training data, and researchers and other people using these models tend «to mistake … performance gains for actual natural language understanding.»

OpenAI Chief Executive Sam Altman acknowledges problems, but he’s pleased overall with the progress shown with GPT-4. «It is more creative than previous models, it hallucinates significantly less, and it is less biased. It can pass a bar exam and score a 5 on several AP exams,» Altman tweeted Tuesday.

One worry about AI is that students will use it to cheat, for example when answering essay questions. It’s a real risk, though some educators actively embrace LLMs as a tool, like search engines and Wikipedia. Plagiarism detection companies are adapting to AI by training their own detection models. One such company, Crossplag, said Wednesday that after testing about 50 documents that GPT-4 generated, «our accuracy rate was above 98.5%.»

OpenAI, Microsoft and Nvidia partnership

OpenAI got a big boost when Microsoft said in February it’s using GPT technology in its Bing search engine, including a chat features similar to ChatGPT. On Tuesday, Microsoft said it’s using GPT-4 for the Bing work. Together, OpenAI and Microsoft pose a major search threat to Google, but Google has its own large language model technology too, including a chatbot called Bard that Google is testing privately.

Also on Tuesday, Google announced it’ll begin limited testing of its own AI technology to boost writing Gmail emails and Google Docs word processing documents. «With your collaborative AI partner you can continue to refine and edit, getting more suggestions as needed,» Google said.

That phrasing mirrors Microsoft’s «co-pilot» positioning of AI technology. Calling it an aid to human-led work is a common stance, given the problems of the technology and the necessity for careful human oversight.

Microsoft uses GPT technology both to evaluate the searches people type into Bing and, in some cases, to offer more elaborate, conversational responses. The results can be much more informative than those of earlier search engines, but the more conversational interface that can be invoked as an option has had problems that make it look unhinged.

To train GPT, OpenAI used Microsoft’s Azure cloud computing service, including thousands of Nvidia’s A100 graphics processing units, or GPUs, yoked together. Azure now can use Nvidia’s new H100 processors, which include specific circuitry to accelerate AI transformer calculations.

AI chatbots everywhere

Another large language model developer, Anthropic, also unveiled an AI chatbot called Claude on Tuesday. The company, which counts Google as an investor, opened a waiting list for Claude.

«Claude is capable of a wide variety of conversational and text processing tasks while maintaining a high degree of reliability and predictability,» Anthropic said in a blog post. «Claude can help with use cases including summarization, search, creative and collaborative writing, Q&A, coding and more.»

It’s one of a growing crowd. Chinese search and tech giant Baidu is working on a chatbot called Ernie Bot. Meta, parent of Facebook and Instagram, consolidated its AI operations into a bigger team and plans to build more generative AI into its products. Even Snapchat is getting in on the game with a GPT-based chatbot called My AI.

Expect more refinements in the future.

«We have had the initial training of GPT-4 done for quite awhile, but it’s taken us a long time and a lot of work to feel ready to release it,» Altman tweeted. «We hope you enjoy it and we really appreciate feedback on its shortcomings.»

Editors’ note: CNET is using an AI engine to create some personalfinance explainers that are edited and fact-checked by our editors. Formore, see this post.

Technologies

GFiber Is Merging With Astound Broadband, Likely Expanding to More Areas

Technologies

Your iPhone Has a Hidden Flight Tracker. Here’s How to Use It

Apple quietly built a real-time flight tracker into iOS.

Flying can introduce an entirely new layer of stress to any trip. Flight delays, cancellations and everything that could go wrong can keep you on edge, so staying up to date with your flight’s status is never far away from your mind.

Luckily, we’re in a world where finding the information for your flight is easily accessible. You can check your airline’s mobile app or even Google your flight number and the latest information is readily available. But did you know there’s a secret way to get your flight information on your iPhone?

The iPhone has had a built-in flight tracker for some time now, but you’d never know it existed if you weren’t specifically looking for it — or searching for the correct terms to pull it up.

Below, we’ll show you how to access the flight tracker so you’re just a tap away from the latest flight stats, giving you a little more peace of mind before your trip.

Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source on Chrome.

How to track your flight via iMessage

Before we start, there are a few prerequisites you must meet:

Make sure iMessage is enabled (it doesn’t work with SMS/MMS).
You’ll need your flight number somewhere in your text messages, whether you’ve sent that information to someone (even yourself) or it’s been sent to you.
The flight number must be sent in this format: [Airline] [Flight number], for example, American Airlines 9707.

Launch the native Messages app on your iPhone and open the text message thread that contains your flight information. You’ll know the flight tracker feature works when the text with the flight information appears underlined, which means it’s actionable and you can tap on it.

If your flight is still several months away or it’s already passed, you might see a message that says, «Flight information unavailable.» You might also see another flight that’s not yours because airlines recycle flight numbers.

You can check your flight status from Spotlight Search, too

If getting your flight information from Messages wasn’t easy enough, you can also grab the details right from your iPhone’s home screen by swiping down and adding your flight number into Spotlight Search. This works with Spotlight Search on your Mac computer, too.

How to access the hidden flight tracker

Although the airline name/flight number format highlighted above is the best way to go, there are other texting options that will lead you to the same result. So let’s say we stick with American Airlines 9707, other options that may bring up the flight tracker include:

AmericanAirlines9707 (no spaces)
AmericanAirlines 9707 (only one space)
AA9707 (airline name is abbreviated and no space)
AA 9707 (abbreviated and space)

I would suggest you keep the airline name spelled out completely and add a space between the two pieces of information — like in the previous section — because for some airlines, these alternative options may not work.

Real-time flight tracking

Once everything is set, tap on the flight information in your text messages. If the feature works correctly, you should see the following two options appear in a quick-action menu:

Preview Flight: View the flight’s details. Tap this to view more information about the flight.
Copy Flight Code: Copy the flight code to your clipboard (in case you want to send your flight details to someone else via text or email).

If you select Preview Flight, at the top of the window, you’ll see the best part of this feature: a real-time flight tracker map. A line will connect the two destinations, and a tiny airplane will move between them, indicating where the flight is at that exact moment.

Underneath the map, you’ll see important flight information:

Airline name and flight number
Flight status (arriving on time, delayed, canceled, etc.)
Terminal and gate numbers (for arrival and departure)
Arrival and departure time
Flight duration
Baggage claim (the number of the baggage carousel)

If you swipe left on the bottom half of the flight tracker, you can switch between flights, but only if there’s a return flight.

For more travel tips, don’t miss our test on whether AI can help you fly more sustainably.

Technologies

Copilot Health Is Microsoft’s Doctor-Built Spin on Medical AI

Microsoft doesn’t want its AI to be your doctor. It wants to make you better prepared when you do see them.

Microsoft is taking a major swing at health AI. The company announced on Thursday that it’s introducing Copilot Health, a new experience inside its chatbot that will bring together all your medical records and wearable data with an AI that’s designed to help you understand it all.

«We are really on the cusp of building a true medical superintelligence,» said Mustafa Suleyman, Microsoft AI CEO. «One that can learn everything about you, all of your health conditions, from your wearable data, your electronic health records, and use that to provide support and insights and intelligence at your fingertips.»

A recent Microsoft survey found that mobile Copilot users ask the chatbot health-related queries more than for any other topic. Copilot Health was built to answer those questions. Microsoft’s health AI was fine-tuned by its in-house clinicians and an external panel of hundreds of clinicians in more than 24 countries. It uses the National Academy of Medicine’s framework for evaluating credible medical sources and information from Harvard Medical School via a 2025 licensing agreement.

Copilot Health is inside the regular, consumer version of Copilot. But it’s an entirely separate experience, designed that way to keep your health information separate from your usual chats. Because it’s been specifically trained for health questions, it ought to be more helpful and accurate than the regular version of Copilot or another chatbot. ChatGPT introduced a similar experience earlier this year.

Your health information won’t pop up in responses from the regular Copilot, only in the new health tab. You can delete your data at any time by simply toggling off a setting — something so easy it raises the question why all AI companies don’t make it that simple to delete your data.

Your information isn’t used to train Microsoft’s AI models, the company says. But your medical information in AI tools like Copilot is not protected under the Health Insurance Portability and Accountability Act (HIPAA).

The benefit of using Copilot Health is having a place where all your medical and health information lives, with an AI that’s trained to help answer your questions about it. You can connect data from your smartwatches and rings, as well as upload your medical records. Through a third-party program called HealthEx, you can upload files from multiple doctors’ offices, hospitals and labs at one time.

Copilot Health is not a doctor

If you choose to share your electronic health record, the AI can make more informed recommendations and reference specific doctors’ visit notes and lab results. But don’t use Copilot Health as a replacement for a physician. What the AI can do is discuss your health concerns, help you prepare for upcoming appointments and help you build healthier habits.

«Copilot Health is not meant to give you a definitive diagnosis or a formal treatment plan, but it’s certainly here to support you in getting to the right answers,» said Dr. Dominic King, vice president of health at Microsoft AI. The former surgeon led the team that built Copilot Health.

For example, it can help you come up with a list of questions to ask your doctor, break down lab results and find a provider that accepts your insurance. Copilot Health can discuss your health concerns, like understanding any new symptoms, but it can’t diagnose or prescribe medication.

Microsoft is doing a slow rollout, beginning with adults (ages 18 plus) in the US, with English as the only language. You can sign up to join the waitlist for Copilot Health now.

There are some existing uses of AI in health care today, but they’re disparate. Wearables have new AI-powered data insights and coaching. Some doctors are using AI scribe tools to take notes during appointments with patients. Administrative and insurance work also has its own AI tools, particularly around claims processing (including making denials, in some cases). The common thread is that none of the AI is without flaws, and it should never be used to make important decisions without human oversight.

For AI believers, the tangled, bureaucratic web of American health care is the perfect place to prove that AI intervention can make a real difference. But AI in health care is like putting a Band-Aid on a gunshot wound — a halfway measure that doesn’t fix the underlying problems.

It’s too soon to tell if Microsoft’s goal of a medical superintelligence is viable. But for now, Copilot Health illustrates a more productive use of AI — more than filling the internet with slop.

«I think it is perhaps the most important and most positively impactful contribution that AI can make in the world,» Suleyman said. «And it’s enormously important to us.»