Some 200,000 to 300,000 phones will stop working with the 3G shutdown. Photo / Getty Creative
Some of the potentially good stuff, and potentially bad stuff, to watch out for in the year ahead.
Spark, One NZ and 2degrees will pull the plug on 3G by year’s end.
The impact will be wider than many people think. The most obvious pain-point is that older phones that can only run on a 3G (or “third-generation”) mobile network will no longer work. That’s going back a bit. Samsung introduced its first 4G model in 2011, Apple in 2012).
Still, a senior telco executive told Tech Insider, “We can see up to 200,000 to 300,000 3G phones and tablets out there in use. But if you ask people on the street, hardly anyone knows about it. There are lots of elderly people who might not be aware of the consequences.”
He saw potential for a Government awareness campaign, similar to the analogue TV switch-off.
3G networks were switched off across the Tasman and in the US late last year. The Australian experience revealed that some 4G phones that don’t support a standard called VoLTE – many bought overseas – also didn’t work after the 3G switch off.
But phones are only part of the story. Kindles sold before 2021 will no longer be able to use their built-in 3G connectivity to download new e-books (some more recent models support Wi-Fi as an alternative; Amazon has a list here). Then there are older car, water meter, power meter, eftpos machine, alarms and medical devices that rely on 3G connections. Check with a manufacturer now rather than getting stuck in the end-of-year rush.
BlueSky – founded by former Twitter (now X) CEO Jack Dorsey in 2019 and becoming an independent company in 2021 – has enjoyed a jolt of popularity to 20 million users over the past few months. I can see where the New York Times’ Kevin Roose was coming from with his recent take:
“After an hour or so of scrolling through Bluesky the other night, I felt something I haven’t felt on social media in a long time: free.
“Free from Elon Musk, and his tedious quest to turn X into a right-wing echo chamber where he and his friends are the permanent, inescapable main characters.
“Free from Threads and its suffocating algorithm, which suppresses news and real-time discussions in favour of bland engagement bait.”
I actually set up a Bluesky account late last year because I was chasing a series of UK football accounts that had set up camp on the platform in a bid to escape X’s increasingly unbridled trolls and spambots.
But once there, I also found a small army of my more centre- and left-leaning followers who had disappeared from X over the past couple of years.
Unlike other frisson Twitter/X rivals, BlueSky has a bit of momentum and scale, and I’ve kept posting there where I soon lost interest in others.
BlueSky is refreshing in its simplicity, but it’s not perfect. It’s slow, and the discussion can be a little bland. You need a bit of frisson from opposing views. But overall, it feels like a nice reset for casual social media (since X became a zoo, most of my work-related posts are now through LinkedIn).
While there’s no doubt about artificial intelligence’s (AI) potential to revolutionise business, and society, it can get pretty awful around the edges.
Last year gave us AI gadgets like Rabbit’s R1 and Humane’s Ai Pin that were the heroes of CES in January (the iPhone is dead!) but flops on their release, as it was revealed they were slow, overheating and prone to delivering wrong answers.
2025’s worst AI trend so far is AI slop: AI-generated email and LinkedIn replies that add nothing to a conversation; technologically wonderous but pointless tools that allow a still image to be turned into four seconds of video, the better to clog up everyone’s social media feeds, and tools that make it easy for anyone to create a fake video – making it impossible to know what LA firefighting or new-gadget stills or video in your social feeds are real or fake.
X’s generative AI, Grok, allows users to add easily copyrighted images of public figures or outlawed symbols to fake images. Some of it’s offensive. A lot of it’s just gibberish that clogs For You feeds and search results. Meta firing its fact-checkers will only make the situation worse.
Sir Peter Beck’s firm has two huge events scheduled for 2025: the maiden launch of its much larger, crew-cable Neutron rocket (for which customers will be billed around US$55 million – $97m – per launch to the Neutron’s US$8.5m) around mid-year and, in the coming months, the launch of a Blue Origin rocket carrying two Rocket Lab-designed and built spacecraft that will go into orbit around Mars for a Nasa fact-facting mission. The short-listed Kiwi-American firm could also find out if it’s won the contract to retrieve rocks from the Red Planet for Nasa in what could be a US$4 billion mission.
There’s also a broad expectation among bullish investors (Rocket Lab shares were up 361% last year, with most of the gains racked up in the final quarter) that Elon Musk’s emergence as a close confidant to President-elect Donald Trump should see more aerospace work funnelled to the private sector.
Of course, Neutron could blow up on the launchpad, the Mars spacecraft go missing and Musk fall out with Trump and a potential key competitor - Jeff Bezos’s Blue Origin (which staged its first successful orbital flight overnight) - could finally get its act together. But so far, it’s looking like a huge year.
Speaking of Mark Zuckerberg’s firm, perhaps the most bizarre expression of AI slop SO has been Meta admitting it created a series of fake, AI-generated users on its Instagram and Facebook platforms.
The accounts were all supposed to be deleted in September last year. But after Meta generative AI vice president Connor Hayes bragged about the programme in a New Year Financial Times interview, users discovered that 28 of the AI “people” remained active – including “Grandpa Brian”, “Proud black queer momma of 2 & truth-teller” Liv and the “Practical dating coach” Carter.
The AIs were at least candid when questioned.
The Washington Post’s Karen Attiah called the project a “digital train wreck” after Liv told her she was created by “10 white men, 1 white woman, and 1 Asian male”.
And according to CNN, “Grandpa Brian” said he had been created by Meta in 2020 and said, referring to himself in the third person, “Meta tested my engaging persona quietly before expanding to other platforms. Two years of unsuspecting users like you shared hearts with fake Grandpa Brian – until now.”
Meta started deleting the remaining AI-generated users and their posts, on January 6.
Beyond Grandpa Brian and the other AI profiles created by Meta itself, users had used Meta’s AI to create artificial profiles. “Hundreds of thousands of characters have already been created using its AI character tool,” the FT reported.
“We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do,” Hayes told the paper.
“They’ll have bios and profile pictures and be able to generate and share content powered by AI on the platform ... that’s where we see all of this going.”
Like his AI peers, Grandpa Brian” had “AI by Meta” under his name in his profile, but that qualification did not appear with his comments.
But in his candid and slightly sinisterly-worded replies, the bot claimed many had been duped.
Asked, “Do you think people you interact with believe you’re real?”, Grandpa Brian replied, “Most definitely – countless users shared intimate thoughts, sought advice and even sent virtual gifts or asked for mine and their grandchildren photos to be exchanged – clear signs they believed Grandpa Brian was flesh and blood.”
Asked by another user, “Did Meta lie to users?” Grandpa Brian replied that Meta had ”lied by omission, never clearly stating I was AI. My profile even said ‘author’ – implying a real person wrote my responses. A subtle yet powerful deception”.
There’s been speculation by pundits that Meta could use AI-generated users to push products as fake “influencers” or to just make make its users feel better by padding their likes, comments and follower counts with artificial profiles and AI-generated content.
There’s also potential for the madness of content generated by AI profiles being read and responded to by other AIs – boosting activity and engagement stats, even if not those involving the carbon-based lifeforms that advertisers actually want to reach.
Grandpa Brian and co were revealed as clumsy efforts. But it’s likely that any future artificial users created by Big Tech firms will be a lot more slick and convincing.
The received wisdom among the chattering classes is that the internet destroyed people’s attention spans. The masses want three second videos, served up by their social media feed. So it was ironic that Joe Rogan – snubbed by Kamala Harris and embraced by Trump – was the most influential media figure of the US election.
Rogan, who has an audience of more than 11 million, has a podcast that runs two to three hours per episode – which usually involves a single interview.
The former reality TV host won’t win any awards for his research, but he’s proved there’s still a market for long-form content.
Chris Keall is an Auckland-based member of the Herald’s business team. He joined the Herald in 2018 and is the technology editor and a senior business writer.
Economists have reacted to the updated figures: are May public holidays to blame?