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

Today’s NYT Strands Hints, Answers and Help for Jan. 9 #677

Here are hints and answers for the NYT Strands puzzle for Jan. 9, No. 677.

Looking for the most recent Strands answer? Click here for our daily Strands hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle, Connections and Connections: Sports Edition puzzles.

Today’s NYT Strands puzzle is a fun topic — but one of the answers describes something I’ve never heard of before. Also, some of the answers are difficult to unscramble, so if you need hints and answers, read on.

I go into depth about the rules for Strands in this story.

If you’re looking for today’s Wordle, Connections and Mini Crossword answers, you can visit CNET’s NYT puzzle hints page.

Read more: NYT Connections Turns 1: These Are the 5 Toughest Puzzles So Far

Hint for today’s Strands puzzle

Today’s Strands theme is: For cubs fans.

If that doesn’t help you, here’s a clue: Like Gentle Ben.

Clue words to unlock in-game hints

Your goal is to find hidden words that fit the puzzle’s theme. If you’re stuck, find any words you can. Every time you find three words of four letters or more, Strands will reveal one of the theme words. These are the words I used to get those hints but any words of four or more letters that you find will work:

LAIR, RAIL, CROP, LAID, BALD, DEAN, BROW, DIZZY

Answers for today’s Strands puzzle

These are the answers that tie into the theme. The goal of the puzzle is to find them all, including the spangram, a theme word that reaches from one side of the puzzle to the other. When you have all of them (I originally thought there were always eight but learned that the number can vary), every letter on the board will be used. Here are the nonspangram answers:

BLACK, BROWN, GRIZZLY, PANDA, POLAR, SPECTACLED (That last one is the one I’ve never heard of! But it’s real!)

Today’s Strands spangram

Today’s Strands spangram is BEARSINMIND. To find it, start with the B that’s five rows to the right and three letters down, then wind up, over and down the whole far-right row.

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

Toughest Strands puzzles

Here are some of the Strands topics I’ve found to be the toughest.

#1: Dated slang. Maybe you didn’t even use this lingo when it was cool. Toughest word: PHAT.

#2: Thar she blows! I guess marine biologists might ace this one. Toughest word: BALEEN or RIGHT.

#3: Off the hook. Again, it helps to know a lot about sea creatures. Sorry, Charlie. Toughest word: BIGEYE or SKIPJACK.

Technologies

Today’s NYT Mini Crossword Answers for Friday, Jan. 9

Here are the answers for The New York Times Mini Crossword for Jan. 9.

Looking for the most recent Mini Crossword answer? Click here for today’s Mini Crossword hints, as well as our daily answers and hints for The New York Times Wordle, Strands, Connections and Connections: Sports Edition puzzles.

Need some help with today’s Mini Crossword? The Across clues were kind of tough today. Read on for all the answers. And if you could use some hints and guidance for daily solving, check out our Mini Crossword tips.

If you’re looking for today’s Wordle, Connections, Connections: Sports Edition and Strands answers, you can visit CNET’s NYT puzzle hints page.

Read more: Tips and Tricks for Solving The New York Times Mini Crossword

Let’s get to those Mini Crossword clues and answers.

Mini across clues and answers

1A clue: Question in a late-night text
Answer: YOUUP

6A clue: Plentiful
Answer: AMPLE

7A clue: Saint ___ and Nevis (Caribbean nation)
Answer: KITTS

8A clue: Baby-bringing bird
Answer: STORK

9A clue: Take care of the tab
Answer: PAY

Mini down clues and answers

1D clue: Himalayan herbivores
Answer: YAKS

2D clue: Fail to include
Answer: OMIT

3D clue: «High five!»
Answer: UPTOP

4D clue: The «U» of UV rays
Answer: ULTRA

5D clue: Annoying to deal with
Answer: PESKY

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

Technologies

I Got Up Close and Personal With Boston Dynamics’ New Atlas Robot

Before Atlas takes its first steps into the world of work later this year, I found myself face-to-face with CES 2026’s most talked-about robot on the show floor.

When I say that I went hands-on with the new Boston Dynamics Atlas robot, I mean that I actually held hands with it. This humanoid robot, which CNET just awarded the Best Robot of CES Award, is one of the most advanced in the world, and I couldn’t pass up the opportunity to get up close and personal with it.

This product version of the robot, which is set to be shipped to Hyundai factories imminently to start working, has been the talk of CES this year. The specific Atlas robot I encountered was a static model that wasn’t turned on or fully operational. Our interactions were, therefore, sadly one-sided. Still, I ran my hands over its soft-touch plastic shell and gently prodded at its finger joints, wondering how it would feel if they gripped me back.

People tend to have varying feelings about humanoid robots — understandable given that they are built to some degree in our image, while also usually being stronger than us, with «brains» that we don’t fully understand. Atlas definitely evokes contradictory emotions for me — even more so when I stood face-to-face with it.

I’m in awe of the engineering, a little fearful of its capabilities, hesitant about what it could mean for the future of humanity and charmed by its design and styling. The periwinkle blue iteration of Atlas that I met on the show floor at CES 2026 almost bears more resemblance to a Dyson product than it does the industrial robots that defined Boston Dynamics’ early days, when it was best known for its work with DARPA.

«There’s a lot of really specific things about this robot that probably look a little weird,» said Zachary Jackowski, Boston Dynamics VP and general manager of Atlas. He pointed to the legs, which he described as «like nothing anyone else was doing.»

Atlas’ thighs are narrow set and in line with the torso, while the calves are wider set, attached to their upper counterparts with a circular joint. This robot is, in fact, all subtle curves and soft lines. There are no harsh edges or stark angles.

During a year when CES has been flooded with humanoid robots, Atlas definitely does stand out due to its design. It appears both less classically human and less industrial than some of its peers, while also lacking the often intimidating, featureless faces they tend to exhibit. Instead, it has two low-set cameras resembling eyes placed where you’d usually expect a mouth to be. Its face is a perfect flat circle, defined by an LED halo that gives it a somewhat Pixar lamp effect.

I asked Jackowski why Boston Dynamics decided to skew so relatively unhuman with this version of its humanoid. «Well, it’s not a human,» he said. «It projects the wrong first impression about a robot to have it pretend to be something that it’s not.»

Particularly in the early days of humanoids, he added, robots won’t have anything like human-like intelligence. People should look at it and see it for what it is — a tool for performing tasks safely and efficiently.

In fact, most of the design decisions were made to keep Atlas as simple, scalable and safe as possible, Jackowski said. I remark that there’s some irony in thinking of a humanoid robot as simple, given the complexity of the technology and development process to bring Atlas to life.

The key to making it simple, Jackowski said, is having a strong enough grasp of the technology to «accomplish the complex thing of building a humanoid robot,» but then being able to take it apart and understand that you can use fewer computers and actuators in it while achieving the same results.

And it’s essential to Boston Dynamics that Atlas is perceived as simple. After all, it’s a general-purpose humanoid, which might eventually be sent far and wide to fulfil all manner of roles. Jackowski calls it the «ultimate generalist.»

Simplicity aside, there are aspects of Atlas that Jackowski believes set it apart from other humanoids at the show. «The repairability of this robot is crazy good,» he said. «The runtime is crazy good. The strength is unlike anything.»

From working in Hyundai’s manufacturing plants, Atlas’s job trajectory is to eventually graduate to many of the same industrial environments where Boston Dynamics’ Spot robot works, before moving to bussing tables in the service industry and eventually into the home. The robot will evolve between now and then, Jackowski said. However, this could be an early glimpse of the type of humanoid that will eventually be our housemate.

That’s some way away, though, which is probably for the best. As I gaze up at Atlas, which I’d guess is around the same height as my husband, my feeling is that, however impressive Atlas is, I’m still not ready for it to move in.