ArduinoMania

“Learn to code.” That three-word pejorative is perpetually on the lips and at the fingertips of internet trolls and tech bros whenever media layoffs are announced. A useless sentiment in its own right, but with the recent advent of code generating AIs, knowing the ins and outs of a programming language like Python could soon be about as useful as knowing how to fluently speak a dead language like Sanskrit. In fact, these genAIs are already helping professional software developers code faster and more effectively by handling much of the programming grunt work.

How coding works

Two of today’s most widely distributed and written coding languages are Java and Python. The former almost single handedly revolutionized cross-platform operation when it was released in the mid-’90s and now drives “everything from smartcards to space vehicles,” according to Java Magazine in 2020 — not to mention Wikipedia’s search function and all of Minecraft. The latter actually predates Java by a few years and serves as the code basis for many modern apps like Dropbox, Spotify and Instagram.

They differ significantly in their operation in that Java needs to be compiled (having its human-readable code translated into computer-executable machine code) before it can run, while Python is an interpreted language which means that its human code is converted into machine code line-by-line as the program executes, enabling it to run without first being compiled. The interpretation method allows code to be more easily written for multiple platforms while compiled code tends to be focused to a specific processor type. Regardless of how they run, the actual code-writing process is nearly identical between the two: somebody has to sit down, crack open a text editor or Integrated Development Environment (IDE) and actually write out all those lines of instruction. And up until recently, that somebody typically was a human.

The “classical programming” writing process of today isn’t that different from the process those of ENIAC, with a software engineer taking a problem, breaking it down into a series of sub-problems, writing code to solve each of those sub-problems in order, and then repeatedly debugging and recompiling the code until it runs. “Automatic programming,” on the other hand, removes the programmer by a degree of separation. Instead of a human writing each line of code individually, the person creates a high-level abstraction of the task for the computer to then generate low level code to address. This differs from “interactive” programming, which allows you to code a program while it is already running.

Today’s conversational AI coding systems, like what we see in Github’s Copilot or OpenAI’s ChatGPT, remove the programmer even further by hiding the coding process behind a veneer of natural language. The programmer tells the AI what they want programmed and how, and the machine can automatically generate the required code.

Building the tools to build the tools allowing any tool to build tools

Among the first of this new breed of conversational coding AIs was Codex, which was developed by OpenAI and released in late 2021. OpenAI had already implemented GPT-3 (precursor to GPT-3.5 that powers BingChat public) by this point, the large language model remarkably adept at mimicking human speech and writing after being trained on billions of words from the public web. The company then fine-tuned that model using 100-plus gigabytes of GitHub data to create Codex. It is capable of generating code in 12 different languages and can translate existing programs between them.

Codex is adept at generating small, simple or repeatable assets, like “a big red button that briefly shakes the screen when clicked” or regular functions like the email address validator on a Google Web Form. But no matter how prolific your prose, you won’t be using it for complex projects like coding a server-side load balancing program — it’s just too complicated an ask.

Google’s DeepMind developed AlphaCode specifically to address such challenges. Like Codex, AlphaCode was first trained on multiple gigabytes of existing GitHub code archives, but was then fed thousands of coding challenges pulled from online programming competitions, like figuring out how many binary strings with a given length don’t contain consecutive zeroes.

To do this, AlphaCode will generate as many as a million code candidates, then reject all but the top 1 percent to pass its test cases. The system will then group the remaining programs based on the similarity of their outputs and sequentially test them until it finds a candidate that successfully solves the given problem. Per a 2022 study published in Science, AlphaCode managed to correctly answer those challenge questions 34 percent of the time (compared to Codex’s single-digit success on the same benchmarks, that’s not bad). DeepMind even entered AlphaCode in a 5,000-competitor online programming contest, where it surpassed nearly 46 percent of the human competitors.

Now even the AI has notes

Just as GPT-3.5 serves as a foundational model for ChatGPT, Codex serves as the basis for GitHub’s Copilot AI. Trained on billions of lines of code assembled from the public web, Copilot offers cloud-based AI-assisted coding autocomplete features through a subscription plugin for the Visual Studio Code, Visual Studio, Neovim, and JetBrains integrated development environments (IDEs).

Initially released as a developer’s preview in June of 2021, Copilot was among the very first coding capable AIs to reach the market. More than a million devs have leveraged the system in the two years since, GitHub's VP of Product Ryan J Salva, told Engadget during a recent interview. With Copilot, users can generate runnable code from natural language text inputs as well as autocomplete commonly repeated code sections and programming functions.

Salva notes that prior to Copilot’s release, GitHub’s previous machine-generated coding suggestions were only accepted by users 14 - 17 percent of the time, “which is fine. It means it was helping developers along.” In the two years since Copilot’s debut, that figure has grown to 35 percent, “and that's netting out to just under half of the amount of code being written [on GitHub] — 46 percent by AI to be exact.”

“[It’s] not a matter of just percentage of code written,” Salva clarified. “It's really about the productivity, the focus, the satisfaction of the developers who are creating.”

As with the outputs of natural language generators like ChatGPT, the code coming from Copilot is largely legible, but like any large language model trained on the open internet, GitHub made sure to incorporate additional safeguards against the system unintentionally producing exploitable code.

“Between when the model produces a suggestion and when that suggestion is presented to the developer,” Salva said, “we at runtime perform … a code quality analysis for the developer, looking for common errors or vulnerabilities in the code like cross-site scripting or path injection.”

That auditing step is meant to improve the quality of recommended code over time rather than monitor or police what the code might be used for. Copilot can help developers create the code that makes up malware, the system won’t prevent it. “We've taken the position that Copilot is there as a tool to help developers produce code,” Salva said, pointing to the numerous White Hat applications for such a system. “Putting a tool like Copilot in their hands … makes them more capable security researchers,” he continued.

As the technology continues to develop, Salva sees generative AI coding to expand far beyond its current technological bounds. That includes “taking a big bet” on conversational AI. “We also see AI-assisted development really percolating up into other parts of the software development life cycle,” he said, like using AI to autonomously repair a CI/CD build errors, patch security vulnerabilities, or have the AI review human-written code.

“Just as we use compilers to produce machine-level code today, I do think they'll eventually get to another layer of abstraction with AI that allows developers to express themselves in a different language,” Salva said. “Maybe it's natural language like English or French, or Korean. And that then gets ‘compiled down’ to something that the machines can understand,” freeing up engineers and developers to focus on the overall growth of the project rather than the nuts and bolts of its construction.

From coders to gabbers

With human decision-making still firmly wedged within the AI programming loop, at least for now, we have little to fear from having software writing software. As Salva noted, computers already do this to a degree when compiling code, and digital gray goos have yet to take over because of it. Instead, the most immediate challenges facing programming AI mirror those of generative AI in general: inherent biases skewing training data, model outputs that violate copyright, and concerns surrounding user data privacy when it comes to training large language models.

GitHub is far from alone in its efforts to build an AI programming buddy. OpenAI’s ChatGPT is capable of generating code — as are the already countless indie variants being built atop the GPT platform. So too is Amazon’s AWS CodeWhisperer system, which provides much of the same autocomplete functionality as Copilot, but optimized for use within the AWS framework. After multiple requests from users, Google incorporated code generation and debugging capabilities into Bard this past April as well, ahead of its ecosystem-wide pivot to embrace AI at I/O 2023 and the release of Codey, Alphabet’s answer to Copilot. We can’t be sure yet what generative coding systems will eventually become or how it might impact the tech industry — we could be looking at the earliest iterations of a transformative democratizing technology, or it could be Clippy for a new generation.

Google's entry-level Pixel phone is already a good value, but it's now a much better deal if you're also looking for wireless audio. Amazon is selling a bundle that combines the Pixel 7a with free Pixel Buds A-Series for the same $499 you'd pay for the phone by itself. The promo doesn't cover the Sea blue model and lasts until July 9th at 2:59AM Eastern, but there are otherwise no catches — this may be just what you need to soundtrack your commute or a long summer walk.

The Pixel 7a is our favorite midrange Android phone, and it's not hard to see why. You're getting nearly the same capabilities as the standard Pixel 7 for $100 less, including the Tensor G2 chip, a smooth 90Hz display and wireless charging. While the cameras aren't quite on par, you'll still get exceptional photo quality that beats some more expensive rivals. Add a healthy OS update policy and long battery life and you might not need anything more. The only catches are the slow wireless power and the lack of storage options beyond 128GB.

The Pixel Buds A-Series, meanwhile, are surprisingly impressive even at their usual price. They provide well-balanced sound, a comfortable fit and tight Google Assistant integration (such as translating conversations). There's no wireless charging or active noise cancellation, but that's expected in this price class. For free, they're virtually unbeatable.

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

The original free-to-play battle royale game Call of Duty: Warzone (now known as Warzone Caldera) will shut down definitively on September 21st. That will allow developers to focus on "future Call of Duty content including the current Warzone (originally called Warzone 2.0) free-to-play experience," Activision wrote in a blog post. 

All gameplay, player progression, inventories and online services will expire on that date. However, any Caldera content purchased in Modern Warfare, Black Ops Cold War or Vanguard will still be accessible in those games.

Warzone came along in 2020, featuring two gameplay modes (battle royale and plunder) plus a single map called Verdansk that supported up to 150 players at a time. It shared a battle pass, weapons and cosmetics with 2019's Modern Warfare. It was an immediate success, reportedly hitting 30 million players just 10 days after launch. 

The game was renamed Call of Duty: Warzone Caledera (after the 2021 Caledera map) following the release of Warzone 2.0 in November of 2022. Call of Duty: Warzone 2.0 was recently renamed simply to Warzone, and Activision is encouraging original Warzone players to jump ship to that title. "Expect a vast amount of gameplay choices across three Battle Royale maps (including Season 04’s new map Vondel), as well as Ranked Play, the DMZ Beta featuring five different Extraction Zones, BlackCell offerings, and more," it wrote.

Some players aren't pleased, though, citing the fact that earlier battle royale games like Blackout from Black Ops 4 still have operating servers, while the original Warzone won't. Others are displeased that resources are being shunted to Warzone: Mobile, and some have pointed out that they'll lose all their Warzone cosmetics, according to Kotaku. 

This episode, Cherlynn is joined by senior reporter Jess Conditt and special guest Michael Fisher to talk about the week of reviews. From the Moto Razr+ to the Pixel Tablet, we look at how these devices fit into our lives and make them better (or worse). Then, we go over the highlights from Summer Games Fest and dig into that Titanic situation.

Listen below or subscribe on your podcast app of choice. If you've got suggestions or topics you'd like covered on the show, be sure to email us or drop a note in the comments! And be sure to check out our other podcasts, the Morning After and Engadget News!

Subscribe!

• iTunes

• Spotify

• Pocket Casts

• Stitcher

• Google Podcasts

Topics

• Moto Razr+ review: a foldable with an external display you’ll actually want to use – 1:56

• Pixel Tablet review: Google made a great smart display and a passable tablet – 31:53

• The doomed OceanGate submarine was piloted with a Logitech game controller – 47:50

• Amazon is shutting down Halo health services at the end of July – 59:15

• Jess Conditt’s takeaways from Summer Game Fest – 1:02:51

• Working on – 1:19:27

• Pop culture picks – 1:20:51

Livestream

Credits
Hosts: Cherlynn Low and Jessica Conditt
Guest: Michael Fisher
Producer: Ben Ellman
Music: Dale North and Terrence O'Brien
Livestream producers: Julio Barrientos
Graphic artists: Luke Brooks and Brian Oh

The July 4th weekend is a time to stand around outside, eating and watching distant fireworks from a backyard. A fire pit is a mighty fine accessory for these festivities, and industry leader Solo Stove just announced a major sale on its line of products to celebrate Independence Day. The sale covers the popular Bonfire 2.0, the extra-large Yukon 2.0, the portable Ranger 2.0 and a whole bunch of bundles and standalone accessories.

This is the best sale to-date for Solo Stove fire pits, beating a previous 45-percent off discount back in May to celebrate that other big summer holiday. The sale extends to sets which typically include a fire pit, a stand, a cover or shield, a handle, a lid and various tools for roasting and grilling over an open flame. You'll find the steepest discounts on the Bonfire Backyard Bundle and the Bonfire Ultimate Bundle, both of which are just about half off. Solo Stove products are consistently well-reviewed, so if you want to stand around an open flame this summer, this might be your best bet.

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

You may be able to pay for purchases and get into train stations without having to physically touch your phone to an NFC terminal in the future. The NFC Forum, which defines the standards for NFC, has revealed a roadmap for key research and plans for near field communication through 2028. Apparently, one of the main priorities for the future of the technology is to increase its range. At the moment, NFC only works if two enabled devices are within 5 millimeters from each other, but the group says it's currently examining ranges that are "four to six times the current operating distance."

That's 30 millimeters or 1.18 inches at most, but it could enable faster transactions and fewer failed ones overall, seeing as a longer range also means there's a lower precision requirement for antenna alignment. In addition, the forum is looking to improve the current NFC wireless charging specification of 1 watt to 3 watts. The capability will bring wireless charging to "new and smaller form factors," the forum said, but didn't give examples of what those form factors could look like. 

Another potential future NFC capability will support several actions with a single tap. Based on the sample use cases the forum listed — point-to-point receipt delivery, loyalty identification and total-journey ticketing — we could be looking at the possibility of being able to validate transit tickets or venue tickets for the whole family with just one tap or a single device. NFC-enabled smartphones could have the power to serve as point-of-sale devices in the future, as well. Apple's Tap to Pay feature already lets iPhone owners use their phones as payment terminals. But a standardized capability would allow more people, especially in developing countries where Android is more prevalent, to use their devices to offer payments for their small businesses and shops. 

These plans are in varying stages of development right now, with some further along than others. The forum doesn't have a clear timeline for their debut yet, but it said that the timeframe for its plans spans two to five years.

It's something that's drilled into you from the first essay you write in school: Always check your sources. Yet, New York attorney Steven Schwartz relied on ChatGPT to find and review them for him — a decision that's led a judge to issue a $5,000 fine to him, his associate Peter LoDuca and their law firm Levidow, Levidow and Oberman, The Guardian reports. Schwartz used it for a case in which a man was suing Colombian airline Avianca alleging he was injured on a flight to New York City. In this case, ChatGPT produced six cases as precedent, such as "Martinez v. Delta Airlines" and "Miller v. United Airlines," that were either inaccurate or simply didn't exist.

In the decision to fine Schwartz and co., Judge P Kevin Castel explained, "Technological advances are commonplace and there is nothing inherently improper about using a reliable artificial intelligence tool for assistance. But existing rules impose a gatekeeping role on attorneys to ensure the accuracy of their filings." Basically, you can use ChatGPT for your work but at least check its claims. In not doing so, the lawyers had "abandoned their responsibilities," including when they stood by the fake statements after the court questioned their legitimacy.

Examples of ChatGPT and other AI chatbots inaccuracies are widespread. Take the National Eating Disorder Association's chatbot that provided people recovering from eating disorders with dieting tips or ChatGPT wrongly accusing a law professor of sexual assault using a non-existent article from The Washington Post as proof. 

Would you watch two billionaires tussle in a cage match? What if it was the owners of Facebook and Tesla? This all started when Elon Musk said he was "up for a cage match if he [Zuckerberg] is," in response to tweets about Meta’s incoming Twitter rival. Musk responded. Mark Zuckerberg posted a screenshot of the exchange as a story on his Instagram account with the note: "Send Me Location." Zuckerberg has been training in Brazilian jiu-jitsu for around a year now, and his efforts have been well documented – he even competed in a tournament back in May and won gold and silver medals. Musk: he said he has a great move called The Walrus. Hmm.

– Mat Smith

The Morning After isn’t just a newsletter – it’s also a daily podcast. Get our daily audio briefings, Monday through Friday, by subscribing right here.

The biggest stories you might have missed

How to make the most of that Instant Pot you just bought

The best iPhone accessories for 2023

Nintendo's new WarioWare game wants you to move your whole body

Meta pulls news content from Canadian Facebook and Instagram

Google's Duet AI can generate custom templates in Sheets

E3 2024 and 2025 aren't canceled (yet)

An LA commission said the in-person shows were done, but the ESA says otherwise.

The Electronic Entertainment Expo hasn't been held in person since 2019. Now, it may not be returning for 2024 or 2025 – at least not at the Los Angeles Convention Center. According to an LA City Tourism Commission planning document shared on ResetEra, the video game trade show has canceled its live event for the next two years.

The Electronic Software Association (ESA) seems hesitant to confirm the entire event is canceled: "ESA is currently in conversation with ESA members and other stakeholders about E3 2024 (and beyond)," the group told Engadget. You can still catch up on everything not-E3 from this year at Summer Game Fest, including our first impressions of Sand Land.

Continue reading.

The next version of Stable Diffusion won’t produce spaghetti hands in AI images

It was a bit jarring.

The next version of the prompt-based AI image generator Stable Diffusion will produce more photorealistic images and be better at, well, making hands look less like a horror show. The announcement appeared in a since-deleted blog post. SDXL 0.9, a follow-up to Stable Diffusion XL, “produces massively improved image and composition detail over its predecessor,” the blog post read. SDXL can be run locally on your PC if you have a powerful enough machine. It requires a minimum of 16GB of RAM and a GeForce RTX 20 (or higher) graphics card with 8GB of VRAM.

Continue reading.

Sony plans to keep making smartphones for at least a few more years

Qualcomm has a deal to power Sony handsets in a multi-year deal.

Sony

You might not be buying them, but Sony will continue to make them. Sony has struck a multi-year deal with Qualcomm to use Snapdragon platforms to power its handsets. This is an extension of an existing agreement. Sony revealed its latest phone, the Xperia 1 V, just last month. The smartphone’s target audience is, once again, photographers and vloggers. Those are relatively niche use cases, though Sony can tap into its camera technology to offer something at least unique. The Xperia 1 V even works as a monitor for compatible Sony Alpha cameras.

Continue reading.

The United States government has reaffirmed its commitment to move EV production to its shores instead of relying on foreign entities. The US Department of Energy's Loan Program Office (LPO) has announced a conditional $9.2 billion loan for BlueOval SK (BOSK) — owned by Ford and South Korean battery producer SK On — to build three battery manufacturing plants, Bloomberg reports. 

The loan is the biggest the LPO has given out yet — almost four times the size of last year's $2.5 billion loan for Ultium Cell — a joint venture between General Motors and LG. The loan's scale is thanks, in part, to last year's passage of the Inflation Reduction Act, which led to the LPO's lending budget increasing to $400 billion. For context, in the previous 14 years, the LPO has dispersed about $33 billion. The extra capital will certainly be necessary to achieve the Biden-Harris administration goal for EVs to make up half of US car sales by 2030. 

There will be two plants in Kentucky and one in Tennessee, with all three producing batteries for Ford and Lincoln's upcoming EV. The car manufacturer also announced plans for a Michigan-based LFP battery plant earlier this year. The production ramp-up comes as Ford aims to roll out two million EVs by 2026, with the All-Electric Explorer, Mustang Mach-E and E-Transit already available and an EV lineup in the works for Lincoln. In comparison, Ford produced about 132,000 EVs in 2022. Ford also recently secured its EV drivers access to 12,000 Tesla's charging points across North America.

The LPO stresses that the loan will also bring career opportunities to the areas, creating 5,000 construction jobs and another 7,500 operation jobs once the plants start running. The investment also aligns with President Biden's Justice40 Initiative that 40 percent of specific federal investments (including LPO loans) go to disadvantaged communities. 

YouTube has plans to go beyond translated subtitles by allowing creators to dub videos into other spoken languages. At VidCon, the company announced yesterday that it's testing an AI-powered dubbing service called Aloud, developed at Google's Area 120 incubator, The Verge reported. The tool would eliminate the time and often great expense required to create a dub the usual way (with human translators and narrators), allowing creators to reach a wider global audience.

Aloud promises a "quality dub in just a few minutes" using AI. The tool first creates a text-based translation that creators can check and edit, then generates a dub. Users can choose different narrators, how to publish and more. Best of all, the service is available for "no charge," Aloud's website states.

YouTube is currently testing the tool with "hundreds" of creators, YouTube's VP of product management, Amjad Hanif, told The Verge. It's currently available in English and lets you dub into Spanish and Portugese with "more languages coming soon," according to Aloud. 

From a user perspective, the setting appears as an "Audio track" toggle in the gear icon, just below subtitles. An example of that is from the Amoeba Sisters science channel trailer, which uses English as a native language, with a dubbed Spanish language option created by Aloud's AI. 

The ability to easily dub languages can expand a creator's reach without the need to do anything else, Google said in its keynote. There's no mention yet as to when Aloud's dubbing tool will be available more widely. In the future, though, YouTube will "make translated audio tracks sound like the creator's voice, with more expression and lip sync," Hanif said.