Technology

If you’re a hot sleeper, gel-infused memory foam could be the solution. Our team hand-picked the best options.

Combine two top kitchen accessories with an air fryer toaster oven from top brands like Breville, Cusinart and more.

Gen Z filers are newer to tax filing, while millennials may find their taxes more complicated after buying property, having children or pursuing side hustles.

Keep your AirPods Max protected and make them personalized with our CNET experts' picks.

Here are the answers for The New York Times Mini Crossword for March 12.

I spend a lot of time poking, prodding, and generally tormenting AI agents to see what makes them tick, and I've developed a keen eye for what’s genuinely useful and what’s merely shiny tech hype.

Recently, my curiosity led me to test ChatGPT's latest feature, Deep Research, and I have to admit – I’m genuinely impressed. But before we jump into why, let’s briefly unpack what exactly this feature is and how it works.

Deep Research is the brilliant librarian addict you always hoped to be paired up with in school. It can gather, analyze, and repackage huge amounts of information in useful and even fun ways. Unlike a typical quick Q&A session, Deep Research takes its time. That can mean anything from five minutes to, theoretically, a full half hour. It will then deliver detailed, trustworthy insights tailored precisely to your query.

I've spent a lot of time with it and genuinely think it's the best bit of ChatGPT right now. If you're unsure whether it's worth the hype, I've put together a handful of reasons you might want to stick a toe in Deep Research.

The perfect vacation

(Image credit: Pixabay)

I've spent more time than I like to think about scrolling through travel blogs, Reddit threads, and outdated TripAdvisor reviews just to plan a simple getaway. Planning vacations, especially if you want something beyond the standard tourist traps, is something Deep Research is quite good at. For a test, I asked the AI to "Plan a two-week trip to Portugal, emphasizing local culture, hidden gems, authentic cuisine, and avoiding crowds."

Deep Research’s response was genuinely impressive. After about 20 minutes, it came back with a carefully planned itinerary filled with interesting recommendations, like a quiet day exploring the historic fishing villages along Portugal’s Silver Coast and sampling fresh seafood in lesser-known local restaurants. It included thoughtful touches, such as attending a small-town festival celebrating traditional crafts, and practical logistics, such as ideal transport options and lodging suggestions outside major tourist hubs.

I gave it another go with, "What’s the best month to visit Costa Rica for wildlife viewing, especially if you’re hoping to see sea turtles?"

Again, Deep Research produced a thorough report. It neatly summarized the nesting and hatching seasons of various turtle species and described weather conditions and tourist patterns month by month. It provided useful tips about where and when to find volunteer opportunities with conservation projects.

Health help

(Image credit: Getty Images)

Deep Research can be a real boon if you've ever had that middle-of-the-night headache and discovered that, according to WebMD, it's a rare tropical disease.

Reliable online health information sometimes takes more than a little digging to work out which bits apply to your hypochondriac moment. Deep Research can personalize your answers while avoiding the more nonsensical suggestions like combatting viruses with crystals. The same goes for more lifestyle-related health queries. For instance, I asked Deep Research to "Investigate whether intermittent fasting has significant long-term health benefits and if it’s safe for everyone."

Instead of sensational headlines, Deep Research returned an extensive, measured overview of clinical studies and expert opinions. It outlined potential benefits like improved metabolism and weight loss. Still, it balanced it with thoughtful cautions about groups who should avoid fasting, such as pregnant women and people with certain chronic health conditions.

Hobbying

(Image credit: Netflix)

Taking on a new hobby can be exciting until you realize the first steps usually involve wading through conflicting advice on YouTube or outdated instructional articles. Here, Deep Research acts like your patient, knowledgeable guide.

I asked the AI to "Create a beginner’s guide for someone who wants to start baking bread from scratch, including common mistakes to avoid."

It took a while, around 15 minutes, to get the report. I presume this is because there was a massive glut of people writing about their experiences with baking in 2020.

Still, I eventually had an accessible yet comprehensive bread-baking guide that even included troubleshooting tips for issues like overly dense dough or uneven rising. It suggested easy starter recipes, explained the science behind yeast fermentation in simple terms, and even included advice on necessary equipment that wouldn’t cost too much.

Purchase power

(Image credit: Shutterstock)

The infinite ocean of online shopping choices can instantly overwhelm the most decisive person. I love the idea of a personal shopping guru who can show their work on what to buy.

So I started by asking Deep Research to, "Compare the top-rated robot vacuums under $300, considering user reviews, battery life, cleaning ability, and ease of maintenance."

Deep Research dove into the specs, scoured customer experiences, and filtered through the noise to deliver a report that helped. It laid out the differences between leading models with real clarity. Within minutes, I knew exactly which vacuum would keep my floors spotless without emptying my wallet. With Deep Research, you might finally get rid of much of that shopping stress.

Nerding out

(Image credit: Andrew_t8 / Pixabay)

We've all spiraled down internet rabbit holes chasing our curiosity, emerging bleary-eyed but fascinated hours later. Deep Research offers a much more efficient and satisfying way to scratch your intellectual itch. It's probably one of the best parts of ChatGPT.

For instance, I was thinking about early printing compared to today's flood of digital content, so I asked Deep Research to "Give an accessible but detailed overview of how the printing press influenced European society and culture in the 15th century."

After half an hour, Deep Research produced an engaging summary highlighting key historical events across the Renaissance and Reformation. It detailed how printed texts democratized knowledge, spurred literacy, and even reshaped politics and religion, all explained clearly enough to satisfy casual history buffs without overwhelming them.

I decided I wanted another flavor of nerdy fun on an intriguing subject, without totally grasping it. I requested that the AI "Simplify and explain quantum computing and its possible impacts in everyday life over the next decade."

After a surprisingly shorter time than the printing press request, ChatGPT had a report that could help me understand things like quantum bits and entanglement. It thoughtfully speculated on near-future applications like drug discovery, cybersecurity, and even predicting traffic patterns, basically delivering a compact TED Talk tailored just for me.

Now, of course, Deep Research isn’t without limitations. It can occasionally stumble, serving up overly confident but incorrect responses. Deep Research’s dependency on publicly available internet content means controversial or rapidly evolving topics may present challenges. But these shortcomings don’t overshadow the immense practical value and ease of use this feature offers

In a world increasingly overloaded with questionable information and digital noise, having a reliable, thorough, and easy-to-use tool like ChatGPT Deep Research genuinely feels like having a brilliant, ever-ready assistant waiting at your beck and call.

You might also like

• I can get answers from ChatGPT, but Deep Research gives me a whole dissertation I'll almost never need
• I tried Deep Research on ChatGPT, and it’s like a super smart but slightly absent-minded librarian from a children’s book
• OpenAI confirms 400 million weekly ChatGPT users - here's 5 great ways to use the world’s most popular AI chatbot

Whether you need fast speeds or budget-friendly plans, these internet providers are our top picks.

The latest 15-inch MacBook Air is bluer and better than ever before—and it dropped in price.

The budget airline also strips other benefits for those paying the lowest fares. How to avoid checked bag fees.

Being active doesn't just add years to your life, it also keeps you feeling your best.

The latest AI assistant to stir excitement is Manus, Latin for hand. An apropos choice since Manus is pitched as the helpful hand for any sort of complex research or content creation.

Manus allegedly outpaces any rivals in sophistication. The most obvious comparison is to DeepSeek, but, while that matchup would be interesting (and one I will probably perform), I think it's worth ranking Manus against the far more widely used ChatGPT.

Manus is invite-only and requires the company to approve your reasons for wanting to test it out. Happily, they've provided a variety of demonstrations showing what Manus can do. Of course, any demo should be taken with a grain of salt since they'll almost certainly show the only examples that portray the AI in the best light. Even so, the demos show what Manus can do.

In fact, some of the demos go well beyond what ChatGPT can accomplish at the consumer level. With that in mind, I stuck to prompts that both models could respond to productively. I also focused on projects the average person might engage with, regardless of their profession. Neither caveat narrowed the options drastically. In the images below, the Manus result is on the right and ChatGPT's output is on the left.

Fantasy Movie

(Image credit: Manus/OpenAI)

I started with a fun movie screenplay prompt: "Develop a fantasy screenplay exploring a hypothetical interaction between Mayan and ancient Egyptian civilizations. Balance historical accuracy with creative liberty. First, compile relevant reference materials. Then, propose potential narrative directions and creative inspirations. Finally, select the most compelling concept and draft a preliminary script outline."

Notably, Manus devised the prompt with only a few concepts then had the AI do research, brainstorm, pick an option, and then produce an outline for the actual script. It came through with a selection of research on the two civilizations, then a handful of seed ideas for the movie before picking the one above and writing a pretty competent outline for a whole film. ChatGPT, while not as eloquent in drafting the outline or deep in its research, nonetheless impressed me with what it came up with. While Manus is more thorough, ChatGPT proved it could match its creativity in some ways.

Adobe Bio

(Image credit: Manus/OpenAI)

Next, I selected a more long-form writing prompt, asking for a whole novel on what would normally be a very dry subject. I asked ChatGPT to follow the Manus prompt, "Write a novel-style biography of Adobe Company, approximately 10,000 words in length. While maintaining factual accuracy, go beyond a simple chronological narrative to create a compelling story."

I'm not saying either turned software development into a gripping narrative, but Manus definitely followed the length demand better. And while I didn't spot any errors in the ChatGPT approach, it felt more like a creatively written Wikipedia article rather than a novel. Manus had a full book, with quotes and lots of context. Perhaps ChatGPT's Deep Research feature could match it, but I wanted to stay focused on the more usual ChatGPT model for now.

Business Cards

(Image credit: Manus/OpenAI)

I was curious about Manus' claim around design and image creation, so I picked the business card prompt for the next ChatGPT comparison, asking it to also, "Create a minimalist, elegant business card for me inspired by Apple's design philosophy. I've attached my resume for reference."

Manus made a very nice business card that does exactly what the prompt said and incorporates the relevant details from the included resume. When I used the same resume and prompt with ChatGPT, the results were somewhat messy. The text isn't all clear (or seemingly in human language), the Apple reference has become much more literal, implying a job at the company, and frankly it's less a business card and more what looks like an actual Apple product, perhaps a personalized mini-battery. Not exactly the professional reminder of yourself you'd want to hand out.

Ocean Meditation

(Image credit: Manus/OpenAI)

For the final test, I went with an interesting writing and audio generation prompt from Manus, "Please help me create a five-minute meditation audio. My name is Wayne, and I like the ocean."

Manus, as expected, made a lovely script, edited it, and had a nice voice read it with accompanying ocean sounds. You can listen to the combination on the demo page, which allows you to adjust the relative volume of voice and wave sounds. ChatGPT wrote a script significantly under five minutes long, and while it produced audio, there were no wave sounds despite the claim in the script. And, even when it's just the audio of the script, it's very obvious which voice will help you meditate. The Manus audio is the first file, and the ChatGPT effort is below it.

Manus magic

After running the Manus prompts on ChatGPT, I can see why the new AI assistant has drawn more comparison to DeepSeek. It does seem to execute prompts in impressive ways. If you're a researcher, a designer, or someone who needs deeply researched lesson plans and multimedia integration, Manus makes sense, if you can get an invite.

That said, ChatGPT is still the best AI for everyday users. If you need an AI that’s quick, reliable, and capable of generating high-quality responses in most circumstances, I'd say ChatGPT (and other AI tools of similar power) has everything you'd need. It gives you what you need without making things harder than they should be. In most cases, ChatGPT is more than enough. But if you want an AI that feels like it belongs in a boardroom or research lab, Manus is probably the right call.

You might also like

• I discovered a surprising difference between DeepSeek and ChatGPT search capabilities
• I pitted Gemini 2.0 Flash against DeepSeek R1, and you might be surprised by the winner
• I can get answers from ChatGPT, but Deep Research gives me a whole dissertation I'll almost never need

Here are hints and answers for the NYT Strands puzzle No. 374 for March 12.

Hints and answers for the NYT Connections: Sports Edition puzzle, No. 170, for March 12.

Hints and answers for Connections for March 12, #640.

Boost your gut health by incorporating these probiotic-rich foods into your diet.

It was easy to miss in Apple's blast of launches last week – which also included the new MacBook Air M4 – but Apple recently launched a new iPad Air M3.

Alongside a fresh processor, the mid-range tablet was given a new, Pro-level Magic Keyboard companion. For such a powerful tablet, it's pricing also remains at a tempting level, starting at $599 / £599 / AU$999 (for the 11-inch model) and $799 / £799 / AU$1,299 (for the 13-inch version).

But should you buy the iPad Air M3? Or is it a pointless middle ground between the refreshed base iPad and the more powerful, OLED-packing iPad Pro M4?

We've rounded up all the biggest reviews from around the internet below to see if there's a consensus and help you decide the best iPad for you. Starting, of course, with TechRadar's experiences with Apple's latest Air...

The iPad Air M3 reviews TechRadar: "Stretches the value even further with more power for the same price"

(Image credit: Future/Jacob Krol) The good

• Power of the M3 chip
• Vibrant screen
• Strong battery life
• Impressive audio

The bad

• No ProMotion display
• Familiar design

Our iPad Air M3 review concludes that it's an even better value option than before – and potentially a reason to skip the iPad Pro (2024). The Air lacks the Pro's best-in-class screen and (ironically) ultra-thin design, but we "really noticed and enjoyed the speed boost" from its M3 chip.

That means, as our review concludes, that the new iPad Air "blurs the line between who this is for and who the iPad Pro is for." It rattled through most tasks we threw at it, from photo and video editing to work apps and gaming. While it isn't far ahead of the M2 chip's performance, we found it "almost impossible" to even slow it down.

In short, the iPad Air M3 is definitely a worthy upgrade "if you have an M1 iPad Air or older, an entry-level iPad, or another tablet" and need more power.

The Verge: "An excellent tablet, and I wish Apple would use iPadOS to make it a more versatile one"

(Image credit: Future/Jacob Krol) The good

• Faster M3 chip
• Big Magic Keyboard upgrade

The bad

• No Face ID
• Apple Intelligence is limited
• Much pricier than base iPad

The Verge's iPad Air M3 review chimes with most of our findings – which means it concludes that it's a great option if you want a slightly more future-proofed tablet than the more affordable base iPad.

Unlike us, The Verge reviewed the 13-inch version. The conclusion was, "if you want a larger model, I’d encourage you to consider the 13-inch Pro instead". We thought the same after crunching the numbers on the iPad Air M3's various configuration options.

Another slight point of difference was that The Verge didn't notice any real-world difference from the M3 chip compared to its predecessor. That means it recommends that "2024 Air plus 2025 Magic Keyboard" could be the combo to go for if you can find it on sale.

CNET: "Once again, the Pro iPad for most"

(Image credit: Future/Jacob Krol) The good

• Speedy M3 processor
• No price rise
• New Magic Keyboard

The bad

• No OLED display
• Can get expensive

Unlike The Verge, CNET doesn't think the iPad Pro is now worth the upgrade compared to the new Air (unless you're considering a high-end Air like the 1TB model).

What's the difference between the two? The Pro gets you an OLED display, thinner design, Face ID camera, and a Lidar scanner alongside an M4 chip. But as we also concluded, the lines between the Air and Pro are now even more blurred – and it depends on how much you want those extra features.

Similarly, CNET's review concluded that there are two big reasons to get the new Air over the base iPad – its M3 chip and compatibility with the Pencil Pro. It also praises the Magic Keyboard, which it says is "a better option now than it was," and in a running theme, says that the 11-inch model "would be the one I'd consider."

Wired: "Now an option for creatives and gamers, but lacks exciting upgrades"

(Image credit: Future/Jacob Krol) The good

• Small performance boost
• New Magic Keyboard case
• 5G connectivity
• Advanced graphics powers

The bad

• Heavier than iPad Pro
• Basic LCD screen
• No new color options
• iPadOS still limited

Wired settled on a middling 7/10 rating for the iPad Air M3, with the main recommendation being to wait for price drops on it if you need the M3 power. "It delivers a great experience, but it doesn't bring anything vital to the table that warrants paying full price," Wireds review concludes.

Its side-by-side comparisons with the M1 and M2 iPad Airs showed that "it was tough to spot a distinction in performance between all three," other than the M3 feeling "a bit snappier" but not noticeably so.

The other main criticisms are the Air's design ("it's comical that the iPad Air is still a few grams heavier and almost a full millimeter thicker than the iPad Pro") plus the lack of battery life and screen improvements. Wired thinks Apple should have replaced its LCD panel with the mini LED one seen on the iPad Pro in 2021 and 2022.

In short, while Wired thinks the Air M3 is still a solid option, it reckons that "we're on the precipice of an even thinner, faster, and better model" in the near future.

Engadget: "A modest update that's still easy to recommend"

(Image credit: Future/Jacob Krol) The good

• Affordable big-screen iPad
• Improved, cheaper keyboard
• Powerful M3 chip

The bad

• Aging screen tech
• No face ID
• Colors could be brighter
• Can get expensive quickly

Like most iPad Air M3 reviews here, Engadget didn't notice much of a real-world impact of the new M3 chip (despite Geekbench scores showing a 16% improvement over the M2). However, it does point out that the chip makes the tablet slightly more future-proofed.

Like Wired, Engadget thinks Apple should have upgraded the Air's screen tech to the iPad Pro's mini LED panel from a few years ago. The combo of a 120Hz refresh rate and 1,000 nits brightness (1,600 nits for HDR content) would, it says, "be a major improvement" given how popular iPads are for watching video. Then again, it also admits that "the iPad Air’s screen remains quite lovely."

In conclusion, Engadget says, "the Air remains a clear upgrade over the base iPad," thanks to its "improved display, stronger performance, more robust multitasking experience and better accessories." But also that most people who want "just a good iPad" likely wouldn't miss the Air's upgrades.

The iPad Air M3 verdict

(Image credit: Future/Jacob Krol)

Despite some minor differences of opinion, there's a clear consensus in the iPad Air M3 reviews – all reviewers think it's a fine mid-range tablet if a slightly boring, incremental upgrade.

The new Air may also potentially be caught in between two stools. The refreshed base iPad remains better value for those who simply want an Apple tablet (and don't care about Apple Intelligence), while the cost of the iPad Air M3 configurations can mount to the point where the iPad Pro M4 might be the better bet (particularly if you need a 13-inch screen).

The main reasons for picking an iPad Air over the base iPad are its M3 chip and Apple Pencil Pro compatibility. The iPad Pro M3, meanwhile, offers a much better OLED display, an M4 chip, Face ID, a Lidar sensor, and better speakers.

But most reviews also concluded that, if the latter aren't important to you, the new Air remains a fine tablet, with the 11-inch model being the sweet spot for size and price.

You might also like

• Wait, what? The brand-new iPad Air isn't out yet and it's already discounted on Amazon
• The OLED iPad Pro is reportedly less popular than expected – and that could mean these changes to Apple's OLED iPad plans
• Apple's new entry-level iPad ups the performance for the same price, but doesn't support Apple Intelligence

Here are hints and the answer for today's Wordle No. 1,362 for Wednesday, March 12.

The update fixes a bug that may have been exploited.

Metallica is coming to Vision Pro. The legendary thrash metal band known for their distinctive melodies and razor-sharp lyrics are now contained within Apple's $3,499 mixed-reality headset.

Vision Pro's latest immersive music experience is Metallica performing hits for a crowd of 65,000 fans in Mexico City. Apple filmed the band last year using 14 "Apple Immersive Video Cameras." While we hoped Apple was referring to the new Blackmagic URSA Cine Immersive, a nearly $30,000 dual-lens 8K camera system built specifically to capture Vision Pro content, Apple let us know it was not that high-end system.

In a release on the experience, Apple says the immersive 180-degree video will put you onstage, giving you "unprecedented access" to band members James Hetfield, Lars Ulrich, Kirk Hammett, and Robert Trujillo.

The cameras were also set up backstage and inside the legendary "Snake Pit," a mosh-pit space for super fans in the middle of the concert stage.

(Image credit: Apple)

Positioning all those 180-degree cameras on and around the stage (some were on pulleys over the stage, some were on remote-controlled dolleys) puts you in the action as Metallica reminds you to "sleep with one eye open" in Sandman and encourages you to "Bang your head against the stage like you never did before" in Whiplash.

They also play the hit One, but the Vision Pro footage does not give you access to the full nearly two-hour concert.

For Metallica and Vision Pro fans, the new immersive video is good news. Still, if you happen to be a Metallica fan without a pricey headset, you can, starting March 14, book a Vision Pro demo at your local Apple Store, where you'll get to watch Metallica perform Whiplash.

Fade to black

(Image credit: Apple)

Apple has been steadily building up its immersive content library with short films like the harrowing Submerged and animations/games like Marvel's What If…? An Immersive Story.

These are compelling showcases of Vision Pro's formidable capabilities. Watching a standard movie about a sinking submarine differs from watching an immersive 3D experience where it feels like you're going under with the sub.

It's also quite memorable to stand almost nose-to-nose with a tightrope walker traversing an expanse between two towering rock edifices or go ice diving with some intrepid Arctic explorers.

Each of these experiences is unforgettable in its own way, but one gets the sense that Apple is still struggling to connect with consumers on these devices.

$3,500 is a significant investment for a headset that will mostly cut you off from those around you, not counting the creepy EyeSight when you wear it.

Apple is clearly deeply invested in the Vision Pro content ecosystem. Still, no amount of compelling content or even Apple Intelligence integration will turn Vision Pro into a must-buy for cash-strapped consumers. Apple probably needs to lower the price and hurry up with the lower-cost Vision Pro alternative.

Nearly two years after Apple first unveiled the Vision Pro headset, it's clear that if Apple doesn't do something, the bell may soon toll for its immersive headset.

You might also like

• Vision Pro at one – I love Apple's revolutionary headset, so ...
• Vision Pro – 8 reasons to buy Apple's next-gen mixed ...
• Apple Intelligence finally arrives on Vision Pro, but it's the ...

Remember when little yellow-clad Marion swaggered into her dad's TV interview, followed by baby bro in a walker? 62 million people watched.