Tech's Biggest Disappointments in 2024: AI Hype, Failed Gadgets, and Ongoing Challenges

Tech's biggest losers in 2024

The tricky thing about naming the year's biggest losers in tech is that in 2024, it once again felt like everyone lost. Amid the depressing spiral that is social media, the will-they-or-won't-they dance of banning TikTok in the US and the neverending edited and deepfaked content that has everyone questioning what's real, the world lost. And it is lost. But a few areas this year stood out as particularly troubling. Specifically, AI and dedicated AI gadgets proliferated more than ever, spreading not only to our digital assistants and search engines but to our wearables as well. We also saw more deterioration in Intel's standing and bid farewell to a robot maker, as well as Lightning cables. I'm pretty happy about that last one, though. Our annual collection of the worst tech developments each year is shorter than usual, but that might be because we're all exhausted. And also because most of the bad things can be attributed to AI, social media or misinformation. Still, we journey down this nightmarish memory lane, hopefully so we can avoid similar pitfalls in future. 2024 was a year in which consumer-facing AI tools became harder and harder to ignore. That's thanks to the tech giants Google, Meta and finally Apple baking AI tools into some of the most-used software on the planet. And in this push to get AI in front of everyone, I cannot help but stop and wonder who exactly is asking for this, and is anyone actually using it? In the past few months, I've been testing a Samsung Chromebook with a host of AI tools built-in as well as trying the various Apple Intelligence features that have rolled out through the autumn. It all came to a head in one of Engadget's Slack channels in early December, just after Apple launched its generative emoji and Image Playground features. Getting Image Playground to spit out AI-created pictures was easy enough, and Genmoji does feel like the logical next step after Apple introduced its personalized Memoji back in 2018. But across the board, the results felt uninspired, off-putting and - -- perhaps worst of all - -- extremely lame. Since I take so many pictures on my iPhone, there are tons of images categorized under my name in the Photos app (it will group together similar faces for years, if you let it). With hundreds of images to pick from, Image Playground should have no problem making a convincing facsimile of me... playing the guitar on the moon, right? Well, yes and no. In this image, as well as ones created of my colleagues Cherlynn Low, Valentina Palladino and Sam Rutherford, there are a few facial characteristics that made me feel that the AI-generated cartoon I was looking at was at the very least inspired by these people. But they all gave off serious uncanny valley vibes; rather than being a cute digital cartoon like we all built with Bitmoji back in the day, these results are soulless representations with no charm and mangled fingers. In a totally different vein, I just had occasion to try out Google's "help me read" summarization features on a 250-page government report. I knew I did not have time to read the entire document and was just curious what AI could do for me here. Turns out, not much. The summary was so brief that it was essentially meaningless -- not unreasonable, as it tried to parse 250 pages into about 100 words. I tried this trick on a review I was writing recently, and it did a much better job of capturing the gist of the article, and it also accurately answered follow-up questions. But given that the final product amounted to maybe four pages, my impression is that AI does a decent job of summarizing things that most people can probably read themselves in the span of five minutes. If you have something more complex, forget it. I could go on -- I've been having a blast laughing at the ridiculous notification summaries I get from Apple Intelligence with my co-workers -- but I think I've made my point. We're in the middle of an AI arms race, where massive companies are desperate to get out ahead of the curve with these products well before they're ready for primetime or even all that useful. And to what end? I don't think any AI company is meaningfully answering a consumer need or finding a way to make people's lives better or easier. They're releasing this stuff because AI is the buzzword of the decade, and to ignore it is to disappoint shareholders. -- Nathan Ingraham, deputy editor This year, no two devices arrived with more manufactured hype than the Humane AI Pin and Rabbit R1. And no two devices were more disappointing either. Both Humane and Rabbit made the argument that people were ready to drop their phones for something smarter and more personal, but neither of their devices were actually good or useful. Of the two, Humane was easily the biggest loser of 2024. The company achieved the ignominious honor of reaching net negative sales because former buyers began returning the AI Pin faster than new units could be sold. I wish I could say its troubles stopped there, but they didn't. After Humane first warned customers that the AI Pin's charging case was a fire risk, it issued a formal recall in October. In the intervening months, the company has reportedly tried to find a buyer without success. Rabbit has certainly faced its own share of troubles, too. After being roundly panned by reviewers in May, a hacker group announced in June that the R1 had huge security holes. In July, it came to light that user chats with the R1 were logged with no option for deleting. Last I checked, Humane has since pivoted to making an operating system that it expects other companies will want to add to their devices, but here's the thing: poor software was a big part of what made the AI Pin bad in the first place. You have to give the company points for trying, but at this point, I would be surprised if Humane is still in business by this time next year. -- Igor Bonifacic, senior reporter This observation has been making the rounds all year long, but if you compare Google from 10 years ago to what it is now, the difference is stark. With the introduction of AI Overviews this year, it felt like Google finally made search results utterly impossible to use without scrolling. Forget sponsored results, newsboxes and discovery panels and all the different modules taking up the top half of the results page for any given query -- in 2024, Google decided to add yet another section above everything, pushing the actual list of websites even further down. Since its initial release in the US in March, AI Overviews told people it was okay to put glue on pizzas or to eat rocks. In spite of the general tendency for AI to get facts wrong, Google continued to expand the feature to more countries, while admitting that the overviews could be "odd, inaccurate or unhelpful." Not only that, it also began to add ads to Overviews, meaning that in addition to the unreliable AI-generated results at the top, people could pay to put what they want to promote in that precious real estate, too. Throw in the fact that the actual results boxes and rankings are all susceptible to SEO gaming by websites trying everything they can to garner a higher spot on the list, and you'll find that Google's search results are basically pay-for-play at this point. And while that will continue to earn the company billions of dollars, it makes finding actually good, high-quality results much more arduous for the discerning user. It gets worse when you consider the priority Google's search engine has on iPhones and Android devices. This year, the US government declared Google a search monopoly, saying the company paid the likes of Apple, Samsung and Mozilla billions of dollars a year to be the default search engine on their devices and browsers. Then there's Chrome, which is the world's most popular browser with its own dubious history around tracking users in Incognito mode. Can we even trust what we see on Google Search any more? People have begun to quit using Google Search altogether, with the rise of alternatives like DuckDuckGo and Kagi, a search engine you'd pay $10 a month to use, as well as OpenAI's SearchGPT, which launched this year. But I'm not convinced that the vast majority of users will switch to these options, especially since one of them costs money and another involves more AI. I can understand that it's hard to make a product that adapts to your users' needs while also keeping your shareholders happy. If only Google (or any big company, really) could re-rank its priorities and bring back a search engine that simply connects people to the best that the internet has to offer. -- Cherlynn Low, deputy editor The road to every great tragedy is paved with people making the most self-serving decisions at the worst possible times. Which brings us neatly to Intel as it burns through its last remaining chances to avoid becoming a business school case study in failure. Earlier this month, it fired CEO Pat Gelsinger halfway through his ambitious plan to save the chip giant from its own worst instincts. Gelsinger was an engineer, brought in to fix a culture too beholdened to finance types who can't see beyond the next quarter. Sadly, despite telling everyone that fixing two decades' worth of corporate fuck-ups would take a while and cost money, it came as a surprise to Intel's board. It ditched Gelsinger, likely because he was trying to take a longer-term view on how to restore the storied manufacturer's success. It's likely the accursed MBA-types will now get their way, flogging off the company's foundry arm, kneecapping its design team in the process. It'll take Intel a decade or more to actually feel the consequences of ignoring Gelsinger's Cassandra-like warnings. But when TSMC reigns alone and we're all paying more for chips, it'll be easy to point to this moment and say this was Intel's last chance to steer out of its own skid. -- Daniel Cooper, senior editor We knew the writing was on the wall when the iPhone 15 debuted with USB-C in 2023, but this year put Lightning's shambling corpse in the grave. The Apple-only connector was a revelation when it debuted in 2012's iPhone 5, replacing the gigantic iPod-era 30-pin connector. Unlike the then-ascendant micro-USB port that dominated Android phones and other small devices in the early 2010s, Lightning was thinner and -- this was key -- reversible, so there was no wrong way to plug it in. It eventually made its way to a large swath of devices in the Apple universe, including AirPods, iPads, Mac accessories and even a Beats product or two. But even Apple relented and started flipping new products to the similarly sized (and likewise reversible) USB-C, albeit years after it had become the dominant standard for data and power connections worldwide. With even holdouts like the AirPods Max and the Mac input devices getting USB-C retrofits in 2024, only a handful of legacy Lightning devices -- the iPhone SE, iPhone 14 and old Apple Pencil -- are left on Apple's virtual shelves, and all will doubtless be gone by this time next year. That's OK: Lightning served us well, but its time has passed. All hail our universal Type-C overlords. So while the death of Lightning is a flat-out win for cross-device charging for the whole world going forward, anyone whose home is still bristling with soon-to-be-replaced Lightning charging stations can be forgiven for feeling a pang of nostalgia in the meantime. -- John Falcone, executive editor When I wrote about Moxie, the child-friendly robot from Embodied, I was charmed by its adorable design and chatty demeanor. It was meant to serve as a companion to children, something that could help them read or simply have conversations. I was less charmed by its $1,499 to $1,699 price, alongside an eventual $60 a month subscription. And now Moxie is officially dead, as Embodied announced it's shutting down operations due to "financial challenges" after a failed funding round. Dead home robots aren't exactly a new phenomenon (remember Jibo?), but Moxie's demise feels particularly rough, since it was a device mainly meant to help kids. Imagine having to tell your child that their robot friend had to shut down because of "financial challenges." Embodied said it would offer customers age appropriate guidance to help discuss the shutdown, but no matter how you spin it, it'll be a tough (and possibly traumatizing) conversation for your youngin. Perhaps it's good to learn early though that all of your smart devices will die. (Not our pets though, they are immortal.) -- Devindra Hardawar, senior editor

8 things that shouldn't have happened last year, but did

Opinion Happy new year! Tradition says that this is when we boldly look forward to what may happen in the 12 months to come. Do you really want to know that? Didn't think so. Instead, here are eight of the tech things that didn't go as planned in 2024, in no particular order and of no particular significance except to paint a picture of the year in tech guaranteed hype free. Why eight. Not five or ten? We just like powers of two. Now, on with the show. This could easily be a mini list all of its own, given the scandalous Java licensing heist. There was a rollout of patient record software in Sweden so shabby people rolled back to manual systems - the tech giant was also treated to nine-digit data privacy fines. The winner has to be HR system so over-budget and under-performing it materially contributed to the bankrupptcy of Europe's biggest local authority. You can't get that level of service just anywhere. Like the British Library, the Internet Archive was attacked this year by cyber criminals causing months of disruption to services many rely on. This is only not repugnant if you don't like the idea of access to knowledge as the basis of a just and free culture. Unhappily for the Archive, this now includes record labels Sony and Universal, who are suing for $600 million in punitive damages over copyright. IA's Wayback Machine, you see, takes historical snapshots of web pages which sometimes include licensed music. Thus, Wayback Machine is starving musicians, which is the record labels' job, and it must be destroyed. That music lawyers find it very useful to counteract claims by music publishers over who did what when is a pure coincidence. If misery loves company, then network outages spread more love around the globe than a Beatles mawkish singalong. One misconfiguration, one botched update. One power supply popping - the butterfly wing flaps that unleash chaos across the planet. One thinks of Instagram's 3.3 million lost connections, or Cloudflare's 5m. Mere runners-up to KING of unreachable - Facebook. With its epic 11.1m user outage. That was all wrapped up in Meta's second big outage of the year, proving Zuck's dedication to providing tech news outlets with enough news to keep going single handedly. (If you're an outage addict, here's the global report for the year.) Like AI and quantum computing, virtual and augmented reality are ideas that seem like good ideas until you try to make them do something useful. It doesn't matter how many billions of dollars are spent trying to prove otherwise. The Apple Vision Pro goggles got a lot of attention at launch. Apple entering a new sector with very expensive tech will never not be news, but since then, LESS BUZZ THAN A FOSSIL BEE IN AMBER. Not only are there no killer apps, there are no apps in development at all. The Apple Lisa did things that nothing else released could do while being too expensive to sell, but the Mac, which was quarter the price, got there eventually. The $3,500 Apple Vision Pro face-hugger at $800 would still be useless. It turns out that despite talk of turning X into a world banking platform, and forcing fleeing advertisers back onto the platform through court action, Elon Musk just wants it as the biggest trolling platform on this or any other planet. He used it to troll England about being close to civil war - a country far too polite to make a fuss - and thumbed his nose at Brazil's attempts to apply its online regulations. Brazil won. Users who like that sort of thing have stayed, but the diaspora of those who don't has left competitor Bluesky with scaling problems. Way to go, Elon. People don't want Copilot PCs and they really didn't want Recall. With a lot of AI, it's hard to pick out any one reason it can fall flat - its intrusiveness, its inability to tell truth from fiction, its insatiable appetite for other people-s data? - but Recall's modus operandi of scraping everything you did on your desktop and sending it to Microsoft is a genuine chart-topper on Planet Creepy. That the company didn't seem to understand this at first, and remains really keen to remain that weird kid with the unsettling stare and no boundaries who wants to be your very best friend no matter how hard you try to avoid them... well, that's not a good sign for sanity in 2025. DARPA kicked off its Grand Challenge long distance autonomous vehicle competition in 2004. As successive years saw huge advances, the ultimate goal of the driverless car felt inevitable. Now, with General Motors unplugging its Cruise driverless taxi project as permanently unviable, leaving just Tesla and Waymo seriously in the game, it looks like just another wrong turn. Twenty years of research and massive advances in sensors, compute, and AI haven't done it. And there's nothing on the roadmap to suggest more time will help. Delivery drivers rejoice: your core skills of dropping parcels in puddles and running away without pressing the doorbell will remain unchallenged by technology for some time to come. "Think of the children" is a real eye-roller of a phrase when used to promote state overreach that can't survive logic. Sometimes it's spot on. Moxie is - was - a super-cute robot for children between five and 10, designed to learn their interests and emotions, and bond with them to teach social interactions. Lots of kids, especially non-neurotypical, will form very strong bonds with a favorite toy. When that toy is an infinitely patient, kindly playmate - that's powerful. But Moxie's core AI was cloud-based, parent company Embodied ran out of cash and turned the servers just like that. And just like that every Moxie in the world was dead forever. Ethical AI is not an empty phrase. So there you are, what wonder we can expect for 2025? Answers on a postcard to Vulture Towers. Or maybe just take the easier option and post your comments in the comments section below.