If you visit any of CNET's AI-written articles, you'll now see an editor's note at the top that says: "We are currently reviewing this story for accuracy. If we find errors, we will update and issue corrections." The publication has added the note after being notified of major errors in at least one of the machine-written financial explainers it had published. 

If you'll recall, CNET editor-in-chief Connie Guglielmo recently admitted that the publication had put out around 75 articles about basic financial topics since November last year. Guglielmo said the website decided to do an experiment to see if AI can truly be used in newsrooms and other information-based services in the coming months and years. Based on Futurism's report, it looks like the answer is: Sure, but the pieces it generates need to thoroughly fact-checked by a human editor. 

Futurism combed through one of the articles Guglielmo highlighted in the post, namely the piece entitled "What Is Compound Interest?", and found a handful of serious errors. While the article has since been corrected, the original version said that "you'll earn $10,300 at the end of the first year" — instead of just $300 — if you deposit $10,000 into an account that earns 3 percent interest compounding annually. The AI also made errors in explaining loan interest rate payments and certificates of deposit or CDs. 

You'll find a huge difference in quality when comparing CNET's articles with machine-written pieces in previous years, which read more like a bunch of facts thrown together rather than coherent stories. As Futurism notes, the errors it found highlight the biggest issue with the current generation of AI text generators: They may be capable of responding in a human-like manner, but they still struggle with sifting out inaccuracies. 

"Models like ChatGPT have a notorious tendency to spew biased, harmful, and factually incorrect content," MIT's Tech Review wrote in a piece examining how Microsoft could use OpenAI's ChatGPT tech with Bing. "They are great at generating slick language that reads as if a human wrote it. But they have no real understanding of what they are generating, and they state both facts and falsehoods with the same high level of confidence." That said, OpenAI recently rolled out an update to ChatGPT meant to "improve accuracy and factuality." 

As for CNET, a spokesperson told Futurism in a statement: "We are actively reviewing all our AI-assisted pieces to make sure no further inaccuracies made it through the editing process, as humans make mistakes, too. We will continue to issue any necessary corrections according to CNET's correction policy."

Apple is already working on a more affordable version of its yet-to-be-announced mixed reality headset, according to Bloomberg and The Information. Based on previous reports about the long-rumored mixed reality device, it will cost around $3,000, which is double the price of the Quest Pro headset Meta released last year. Apple's engineers are reportedly working to develop a more budget-friendly version that's closer to the price of the Quest Pro and are already discussing strategies to achieve that goal, such as using cheaper components. 

While the flagship device will have 4K internal displays for each eye, its more affordable counterpart will likely use ones with lower resolution. The company's engineers are also considering the use of fewer cameras and slower processors, which would, in turn, eliminate the need for internal fans. Apple might ask users to manually adjust their field of view for the cheaper headset instead of equipping it with a motor that can automatically do that. It might choose not to equip the device with its custom H2 wireless chip that would allow it to work better with AirPods, as well. A Bloomberg report earlier this month claimed the tech giant is developing its own Bluetooth and WiFi chip to replace Broadcom's by 2025. The Information said Apple could also choose to use that in-house chip to prevent a third-party company from affecting the device's pricing.

The cheaper mixed reality headset is in its very early stages, the sources reported, and Apple has yet to build a working prototype. According to Bloomberg, though, the majority of Apple's 1,000-person Technology Development Group is working on the two headsets and that the company could release it next year or in 2025. As a result of this dual-device strategy, though, Apple has reportedly put its plans to launch lightweight augmented reality glasses on hold. The company's original vision was to create AR glasses that could one day replace the iPhone, but it has now scaled back work on the project. It's now apparently unclear if the AR glasses will ever launch. 

As for Apple's flagship mixed reality headset, Bloomberg previously said that the company plans to formally announce it this spring ahead of WWDC in June. 

If you're looking for new routers, mesh WiFi systems and smart home devices, such as lights and plugs, TP-Link is selling quite a few at a discount on Amazon. The TP-Link Archer AX73 WiFi 6 router is currently available for $140, which, while not quite its lowest price on the website, is still 30 percent off retail. The company says it was designed with 8K streaming in mind. If you're a frequent and serious gamer, TP-Link's Archer GX90 WiFi 6 Gaming router is also on sale for $200, which is the lowest we've seen it go for on the website. 

Archer GX90 is a tri-band router, which provides a 4.8 Gbps dedicated band for your gaming needs. It can detect and optimize game streams, and it comes with eight detachable antennas that give it the power to provide WiFi coverage for a four-bedroom house. Meanwhile, TP-Link's Archer AXE300 Quad-Band WiFi 6E router has dropped back down to its all-time low of $480. It can deliver WiFi speeds of up to 15.6 Gbps and provide access to a new 6 GHz band. Like the Archer GX90, this router also comes with eight antennas to ensure WiFi coverage for big homes. But if you still need mesh WiFi systems to help boost coverage, there are a couple on sale right now, as well. 

You can get a three-pack bundle of the TP-Link Deco P9 Hybrid Mesh WiFi system, which can provide coverage for a 6,000 sq. ft. home, for $150. That's 35 percent less than its typical price. Meanwhile, a two-pack bundle of the TP-Link Deco X20 is on sale for $104 or 42 percent lower than retail. The two WiFi 6-capable mesh devices can cover over an area up to 4,000 sq. ft. in size. 

If you're looking for smart home devices, you can also find Kasa and Tapo by TP-Link-branded products in the list. The Kasa Smart Plug Mini is on sale for $13, down $7 from retail. It can add voice control support to any outlet, particularly for Alexa and Google Assistant. The device also allows you to switch off electronics from anywhere using the Kasa app and to monitor the consumption of anything that's plugged into it. In addition, the TP-Link sale includes Kasa Smart Bulbs for $35 (22 percent off), as well as Kasa and Tapo light strips for up to 44 percent off. 

Follow @EngadgetDeals on Twitter and subscribe to the Engadget Deals newsletter for the latest tech deals and buying advice.

Disney+ has released a new trailer for The Mandalorian during the NFL Wild Card Game on ESPN and ABC Network. It shows Pedro Pascal's character Din Djarin and Grogu reunited and going on their next adventure. Din also reveals in the trailer that he's going back to the planet Mandalore in an effort to redeem himself after removing his helmet and showing his face to other people by the end of season 2. As revealed in previous episodes, Din is a member of the religious sect Children of the Watch that views removing one's helmet in the presence of others a serious transgression.

While Din is dealing with the consequences of his decision, the New Republic is struggling: "There's something dangerous happening out there," Captain Carson Teva warned. "And by the time it becomes big enough for you to act, it'll be too late." The trailer also shows Grogu exhibiting better control of the Force after leaving with Luke Skywalker in the previous season to train at his Jedi Temple. 

Disney released its first teaser trailer for the show at last year's D23 Expo, showing us that everybody's favorite Star Wars family will indeed get reunited for season 3. Now this newer trailer gives us a taste of what Din and Grogu will encounter. Not that we have long to wait for the next season to drop — season 3 will be available for streaming on Disney+ starting on March 1st.

Microsoft is giving more people — or at least more customers — access to OpenAI's technologies, including ChatGPT. The tech giant has announced that it's now making the Azure OpenAI Service generally available after giving a limited number of enterprise customers access to it when it debuted in November 2021. As Bloomberg notes, customers who have access to the service can use various OpenAI tools for their own cloud applications, including the Dall-E AI art generator and the GPT-3.5 language system. Microsoft says it's also adding access to ChatGPT, which it describes as a "fine-tuned version of GPT-3.5," to the service "soon."

The tech giant has been associated with OpenAI ever since it invested $1 billion in the Elon Musk-founded startup back in 2019. This announcement comes shortly after reports were published that Microsoft is in talks to invest an additional $10 billion in the company. "These [AI] models are going to change the way that people interact with computers," Microsoft's head of AI platforms Eric Boyd previously told The Financial Times in an interview.

While OpenAI has been around for a while, it was recently thrust into the spotlight following ChatGPT's debut. The program has the ability to return long, coherent answers that aren't immediately recognizable as machine-generated responses. It was good enough to alarm educators, who expressed concerns that it could be used for cheating. Earlier this month, New York City public schools banned ChatGPT from school devices and WiFi networks.

The Information also previously reported that Microsoft was planning to integrate the OpenAI software powering ChatGPT into Bing. While it's still unclear what the software could do for the search engine, sources said it could enable Bing to return results in a format that's friendlier and easier to digest. ChatGPT is available for free at the moment, but OpenAI intends to make money off it in the future and has already opened a waitlist for those interested in testing a paid version of the bot.

On January 11th, the Federal Aviation Administration paused all domestic departures in the US after its Notice to Air Missions (NOTAM) system failed. The agency later revealed that the outage was caused by a database file that was damaged by "personnel who failed to follow procedures." Now, according to a new report from The Washington Post, the database failure also created issues for tools used by US military pilots. 

One of the affected systems was the Defense Internet NOTAM Service (DINS), which typically comes with FAA alerts regarding flight hazards. During the outage, military pilots were either getting NOTAMs in duplicates or not getting any at all. The Post said an FAA bulletin notified military users that the system had become "impaired and unreliable." Unlike civilian flights, which had to be grounded, military flights can proceed in situations like this. An Air Force spokesperson told the outlet that the military branch's pilots had to call around to ask for potential flight hazards themselves. 

The outage had also erased all NOTAMs submitted to the system starting on Tuesday afternoon, so airports and air traffic controllers were asked to re-submit them. Further, the FAA had to deal with delays and other challenges after the system went back up due to a "high system load."

The FAA is still verifying what caused the outage, but The Post said it's looking like the contractors truly made mistake and that there was no malicious intent behind their actions. Lawmakers are using this opportunity to put a spotlight on the FAA's outdated technology and to seek funding for upgrades. The computer system that failed and led to the outage is already three decades old, and according to CNN, it's also at least six years away from getting an upgrade. It remains to be seen if the incident will change that timeline.

Twitter is now accepting signups for those who to be among the first to access the verification for organizations program. It was previously known as Blue for Business, the company said in its announcement, along with a link to the sign-up form. Organizations will have to submit their names, Twitter usernames and websites to be considered for the waitlist. They also have to indicate their size and the expected number of affiliated accounts. If you'll recall, Musk previously announced that the website will roll out a feature that will give organizations the capability to identify accounts that are actually associated with them.

His announcement came after a rather disastrous launch of Twitter's paid verification system, which gave rise to a bunch of verified trolls impersonating companies, celebrities and other high-profile personalities. This upcoming feature is meant to help address the issue and ensure that users claiming to be part of a specific organization are who they say they are. 

Musk also announced back then that the website will offer checkmarks in different colors: gold for companies, grey for government and blue for individuals. That will make posing as a company or a government agency difficult for random users. The company ultimately had to pause Blue's initial rollout due to the influx of impersonators before relaunching it in December with an $11-per-month price tag. 

Twitter will publish its "tweet recommendation code" and will make tweet and account status visible "no later than next month," Musk has revealed, as well. Presumably, that means users will know if they've been shadowbanned and their tweets aren't showing up for other people. "Transparency builds trust," he added. In addition, he announced that the website is moving the bookmark button to the tweet details page and is fixing its image auto-cropping feature next week.

Several third-party clients for Twitter are having issues communicating with the social network, leading to issues that prevent users from being able to log in. As TechCrunch reports, Tweetbot and Twitterific have both confirmed that they're having problems and are trying to find the root cause of the issue. "We've reached out to Twitter for more details, but haven't heard back," Tweetbot announced. 

Fenix has also confirmed that its client for Android is experiencing problems but that its iOS app seems to be unaffected. Matteo Villa, the app's developer, said Fenix for Android was suspended with no communication from the company. Villa tweeted a screenshot showing a notification that says Fenix "violated Twitter Rules and policies" and can no longer be accessed as a result. The developer has temporarily pulled the app from Google Play while trying to determine the cause. 

Twitpane has posted an announcement in Japanese that says users have been getting errors, such as "account authentication failed," that prevent them from being able to access the app. Other clients that seem to be experiencing issues include Echofon, Feather and Talon. 

Twitter has yet to issue an announcement through any of its official accounts. On the Twitter Developer forum, a poster said users have been encountering a lot of bugs with the Twitter API for third party apps since December. It's unclear what exactly has been preventing third-party apps from accessing the social network, and since Twitter no longer has a communications team, even the affected developers have yet to hear from the company directly.

Tim Cook is getting paid around 40 percent lower than last year, according to the annual proxy statement (PDF) Apple has released — and the CEO himself recommended it. Apparently, during the tech giant's annual advisory meeting for shareholders in 2022, only 64 percent of the "Say on Pay" votes cast regarding the compensation proposals for executives was in favor of retaining their 2021 pay packages. While that's still majority of the votes, it represents a significant year-over-year decline in approval. As 9to5Mac notes, 94.9 percent of the shareholders who voted the previous year was in favor of the compensation proposals for executives. 

To decide this year's pay packages, Apple's Compensation Committee took into account the Say on Pay's voting results and Tim Cook's own recommendation "to adjust his compensation in light of feedback received." For 2023, Cook's target salary is $49 million, down $35 million from his target salary in 2022. His base pay is still $3 million and his annual cash incentive remains unchanged at $6 million, but his equity award value went from $75 million in 2022 to $40 million this year. Further, he was granted an equity award that's 75 percent performance and 25 percent time-based vesting, instead of 50-50 like 2022's. 

The truth is that Cook, who vowed to donate his fortune to charity a few years ago, will likely earn more than $49 million this year due to stock awards and bonuses. According to Bloomberg, he earned $99.4 million in 2022, which was $15.4 million more than his target salary for the year. In 2021, his total pay package was $98.7 million. Critics like advisory firm Institutional Shareholder Services previously urged shareholders to vote against Cook's pay package, citing concerns about how big his equity award is and how it's structured. "Half of the award lacks performance criteria," the firm previously said. This shift in Cook's compensation reflects the changing attitude towards executive pay, and the CEO might be setting an example for his peers. It is, after all, uncommon for an executive his level to recommend that their own pay be cut. 

In a 2022 report about VR fitness, The Washington Post cited a Forrester research that said 25 percent of American adults online are interested in buying a VR headset and 18 percent think they'd primarily use it for exercise. VR fitness is a growing space, and Meta has rolled out a couple of new health-related features for its Quest headset that could convince even more people to give it a try. In its latest update for the device, the company has rolled out Health Connect integration for Android users, and, as The Verge reports, the ability to pair Bluetooth heart rate monitors with it. 

A Meta spokesperson told the publication that if paired with a compatible heart rate monitor, the Meta Quest Move overlay can display a user's heart rate on any game or app, alongside minutes exercised and calories burned. At the moment, the only known compatible monitors are the Garmin HRM-Dual and the Polar H10 chest straps, but other brands and models (even smartwatches) may be able to pair with the headset. Meta just can't confirm if they'll work with the Quest as it intended. 

Meanwhile, Health Connect integration will give users the capability to automatically sync VR workouts with their Android device. Meta first made people's VR fitness stats viewable outside of their headsets when it gave iOS users the capability to sync their progress with Apple Health last year. Now, Android users can finally see their progress without having to put on the Quest, as well. 

The Health Connect API enables health and fitness data sharing across devices and between apps, such as Samsung Health and MyFitnessPal. To activate the integration, users will first have to sync their Meta Quest Move stats with the Meta Quest mobile app. They can do so by going into the Move app in VR and clicking Settings. Users will then have to toggle on "Health Connect by Android" under the Connected Apps section inside the Meta Quest mobile app.

It's unclear if these features are also making their way to the first Quest model released under the Oculus branding. A few days ago, the company sent out emails to Quest 1 owners, telling them that their devices will no longer be getting new features going forward. Meta will also only be releasing critical security and bug patches for the device until 2024.