Start Up No.1927: ChatGPT’s dating failure, early Wired UK, whither podcasting?, vaccines for bees!, scamming scammers, and more

New data from Google says that there are now 150 million devices running Google/Android TV. Is that a big number in context, though? CC-licensed photo by Brian Bilek on Flickr.

You can sign up to receive each day’s Start Up post by email. You’ll need to click a confirmation link, so no spam.

Mark your calendar for Friday: there’s another post due at the Social Warming Substack at about 0845 UK time.

A selection of 11 links for you. Nourishing. I’m @charlesarthur on Twitter. Observations and links welcome.

ChatGPT wrote my dating app responses. No one replied • Business Insider

Jordan Parker Erb:


The program can be used to cheat and deceive. But can it be used to find love? I decided to put it to the test.

I fed ChatGPT a few comments that were left on my Hinge profile, and also asked it to come up with responses to other people’s prompts. Some of the responses were so unbearably cringe-worthy that I gave myself the ick. Others read like a Buzzfeed-era millennial wrote them circa 2010.

To put it kindly: ChatGPT is not a smooth talker.

One person said their most irrational fear was flying. I asked the chatbot to come up with a funny reply, to which it said: “No problem, I’m more than happy to hold your hand and provide moral support during turbulence. And if the plane goes down, at least we’ll go out together in a romantic blaze of glory!”

I sighed. I hit send. I never heard back.

In reply to a flame emoji left on one of my photos, ChatGPT offered: “Thanks for the hot compliment! I’m flattered, but I’m not sure if I can handle all that fire. Maybe we can start with some cozy cuddles and a nice glass of wine instead?”

In another prompt, I asked the bot to write a short, witty response to someone whose profile says they work in finance. In response, it gave me the most deranged and clearly robotic response yet.

“Hey there finance person! I see you’re good with numbers. Can you help me with my budget? I’m trying to save up for a lifetime supply of avocado toast and craft beer. Is that a good investment?”

The “finance person” ghosted. Understandable.


So we’re already at the “try absolutely anything with this new toy” stage. What comes next?
unique link to this extract

Wired UK : a limey whinges • Spesh


IIn the hope that it will somehow protect me from my fellow Haddocker’s undying hatred for what I did back then, some historical perspective: this was 1994. Delphi had just been bought by Murdoch. HotWired was a month or so old. The Electronic Telegraph was there, just about. Most people when asked had not heard of Bill Gates. We didn’t know it, but these were prehistoric times: the crypto-Mosaic era.

So, I’m sitting in Louis Rosetto’s brand new giant office in SOMA with Louis, Kevin Kelly, Jane Metcalfe, and John Plunkett (Wired’s designer). Kelly asks me what the UK scene is like. And, I’m thinking “well, it’s Cix, and it’s demon.local, but I fucking hate them because they’re brain-damaged jabbering fools who think it’s the height of sophistication to express their crippled emotional needs in terms of Blackadder quotes and I’m fucked if I’m giving this to them”. (You know what I mean.)

So I said, ‘I think the UK has to grow its own scene – I think people are waiting for something like this to happen there. But it will be similar – there is a global coincidence of desires for this’. God, I was proud of that ad-libbed phrase: a global coincidence of desires.

Kevin Kelly was really kind – I only ever met him again a couple of times. He didn’t seem to spend much time at Wired. He just popped in occasionally with a new sparkling idea: a self-editing Negroponte. Jane Metcalfe was chatty, and explained about how they’d employed so many new people, and how amazing it was that so many of them were Scorpios(?), and hugged me and hoped that I was well. Plunkett was a smiling happy man of whom, for some reason, people were apprehensive.

Not as apprehensive, I have to say, as they were of Louis.

I was scared of Louis from the moment he entered the room. And as he glanced around at us, it was clear the start that he had rumbled me too.

I was not, his long stare seemed to indicate, nor would I ever be as long as I lived, one of the Wired.

After that meeting, I had a one-to-one with John Battelle, the editor of Wired US. Battelle is a pumped-iron, testosterone-bristled Hemingway of a geek. Que efficient! He asked me how Wired UK was going to run. It was at this point that I, if it was I, made the “decision”, if it was a decision, that I think fucked Wired UK from the start.

I said, “Well, obviously, we’ll create some of our own content, take the best of your articles, and wrap them into one magazine”.


(I can’t figure out who the author was.) Obviously this refers to the original Wired UK, not the version that you see now.
unique link to this extract

GPTZero • Streamlit

This is an app that’s meant to detect whether text came from ChatGPT (or siblings). I tried it on the extract about Wired: this is by a human! I tried it on the two chatup lines from ChatGPT: this is by a human!

Needs refinement, maybe, but a fair first effort.
unique link to this extract

Google TV and Android TV have 150 million devices as users grow • 9to5 Google

Ben Schoon:


Google says that Android TV OS – the underlying platform for both the Google TV and Android TV experiences – now has 150 million monthly active devices. That total isn’t just for Google TV devices or for Android TV devices, but all of them combined. It also doesn’t directly convert to a user total, as some may own more than one Google TV device.

Regardless, this is solid growth over the past year.

As of January 2022, Google had said that Android TV OS had 110 million monthly active users. Year over year, that’s roughly 35% growth. Last time Google shared this stat, it had managed 30 million new devices over the course of seven or so months, while it took a year to hit 40 million this time around.


I’m amazed the number is so small. Google TV and Android TV have been incorporated into products for more than a decade. This implies only about 10 million devices sold annually in preceding years. Or, perhaps, quite a lot of turnover where people buy a capable device, and then replace it a few years later, so that user base doesn’t get broader.

For comparison, Android (on phones and tablets) got to a billion in less than six years.
unique link to this extract

2022: the year that podcasting died • Medium

Nick Hilton is a writer and podcast entrepreneur:


Podcasting is at a precarious point in its existence. Champions of the format will point to slightly ludicrous research suggesting that 60%+ of Americans listen to podcasts. But the figure for actual, current listeners is probably closer to 30%. Comparatively, about 90% of people still listen to radio. So whilst podcasting has undoubtedly entered the mainstream, it still has nothing like the penetration of the core media formats: radio (90% cut-through), TV (80%), books (73%), music (90%). What I think this means is that that success is undoable. With television, say, or music, the genie is out the bottle; the toothpaste out the tube. But with just 30% market penetration, that cut-through is reversible, especially given how many non-podcast mediums offer that same podcast experience. If you start watching Lex Fridman videos on YouTube or listening to Twitter Spaces with Dave Ramsey, the content your consuming hasn’t really changed. The content survives, the distribution mechanism changes.

But this is not really anything to worry about.

Whether you’re a content creator or a technologist, the coming months and years will be about going with the flow.


Look, there isn’t the faintest chance I’m going to watch a Lex Fridman video. The guy read 1984 and produced a book review video on it. An hour-long one. Nobody’s that interesting. Well, he certainly isn’t.
unique link to this extract

USDA approves first vaccine for honeybees • The New York Times

Remy Tumin:


A biotech company in Georgia has received conditional approval from the US Department of Agriculture for the first vaccine for honeybees, a move scientists say could help pave the way for controlling a range of viruses and pests that have decimated the global population. It is the first vaccine approved for any insect in the United States.

The company, Dalan Animal Health, which is based in Athens, Georgia, developed a prophylactic vaccine that protects honeybees from American foulbrood, an aggressive bacterium that can spread quickly from hive to hive. Previous treatments included burning infected colonies and all of the associated equipment, or using antibiotics. Diamond Animal Health, a manufacturer that is collaborating with Dalan, holds the conditional license.

Dalail Freitak, an associate professor in honeybee research at the Karl-Franzens University of Graz in Austria and chief science officer for Dalan, said the vaccine could help change the way scientists approach animal health.

“There are millions of beehives all over the world, and they don’t have a good health care system compared to other animals,” she said. “Now we have the tools to improve their resistance against diseases.”


This is remarkable. Perhaps we’re finally, at long last, going to save the bees. (It’s administered as food, not eeny-weeny injections.)
unique link to this extract

A celebrity photo lawsuit claims Twitter’s copyright process is broken • The Verge

Adi Robertson:


As Twitter slashes staff and pares back moderation under Elon Musk, it may soon run into a problem it can’t ignore: its potential liability for copyright infringement. TorrentFreak reported last week that the company faces a lawsuit by celebrity photo agency Backgrid. Backgrid says that Twitter systematically failed to take down unauthorized copies of its pictures, arguing it shouldn’t be protected by American safe harbor laws for copyright. The allegations leave plenty of questions unanswered, but claims like them could pose a problem for Musk’s increasingly bare-bones operation.

Backgrid’s claims predate Musk’s ownership of Twitter, dating to at least September 2021. The complaint, filed in a California court in late December, alleges Backgrid sent thousands of takedown notices for photos. “Not a single work was taken down and not a single repeat infringer was suspended,” it says. It includes links to some images that remain online as of today, although others appear to have been deleted. It also says it tried to resolve the issue with Twitter before filing the suit, “but Twitter did not respond.” It’s not clear whether this attempt came before or after Musk’s acquisition; an attorney at the firm representing Backgrid did not reply to a request for comment, and Twitter no longer has a public relations department.

…Rebecca Tushnet, an intellectual property expert and professor at Harvard Law School, says the DMCA’s protections are flexible, and Backgrid’s suit only tells one side of the story. “It’s really hard to tell from a complaint what’s going on. One possibility is that the notices were defective in some way. One possibility is that there’s been a maintenance failure and the removals just didn’t get done,” she says. “People find exciting new ways to screw things up all the time, so I don’t want to make any predictions.”


Possible that this is nothing, possible that it’s going to really going to make a hole in Twitter’s finances.
unique link to this extract

Vigilantes for views: the YouTube pranksters harassing suspected scam callers in India • Rest of World

Andrew Deck and Raksha Kumar:


Many of the YouTube creators who make scambaiting videos are from North America and Europe, and their most frequent targets are in India. Oftentimes, scambaiters simply annoy scammers: They might pretend they are falling for a scam call, for instance, only to waste the caller’s time with inane questions or inside jokes. In these videos, the scammers usually remain nameless and faceless, just a voice on the other end of the line. 

But Trilogy has taken things up a notch. In April 2022, their team traveled from Los Angeles to Kolkata in order to prank workers at Ansh Info Solutions and two other call centers, which they claim conduct scam call operations that allegedly defraud victims in the U.S. and elsewhere. Naturally, they filmed the whole thing, hoping to pull it together into their newest viral video.

Kulik and Bingham say their goal is to educate viewers about scams — a public service disguised as comedic entertainment. They talk about how they’re motivated by a sense of justice for victims of scams, and suggest that they’re stepping in where law enforcement has failed.

But there are other advantages to being a scambaiting creator. In the month after the Kolkata videos were released, Trilogy Media’s YouTube channel added more than 140,000 subscribers, according to YouTube analytics site Social Blade. The videos from their trip, posted across their channel and those of several collaborators, collectively have over 60 million views.


As the article notes, this is one of those moral questions: is it bad to do bad things to people who you think are doing bad things? I was writing about this class of scammers back in 2010, and they weren’t new to the scene then. The police in Kolkata aren’t that great at prosecuting them. But they’ve probably got more local problems to deal with.
unique link to this extract

Will the metaverse be your new workplace? • BBC News

Jane Wakefield:


When we look back in 50 years’ time, it is likely that the 2D internet we now all use will seem laughably archaic.

Not only will the internet likely no longer exist behind a screen, but it is probable that we will interact with it differently. We’ll manipulate objects using augmented reality (AR), explore virtual-reality (VR) worlds, and meld the real and the digital in ways we can currently not imagine.

And what will that mean for the world of work? We are already transitioning away from the nine-to-five commute, and turning our backs on the traditional office setting. This is thanks to two years of pandemic lockdowns, and a newfound love of, or tolerance for, virtual meetings.

So will the logical next step be working in the metaverse, the planned virtual universe where cartoon-like 3D representations of everyone will walk around, and talk and interact with others?

The metaverse has become an over-hyped term, so it’s important to note that it doesn’t actually yet exist. And even those invested in the concept disagree about exactly what it will be.

Will rival virtual worlds interconnect in a way that simply doesn’t happen at the moment between competing technologies? Will we spend more time there than in the real world? Will we need entirely new rules to govern these new spaces?

None of these questions have answers yet, but that hasn’t stopped a barrage of interest and hyperbole as firms see a new way of making money.


Back in 1998 I was shown a “3D internet shopping mall”, offered on the basis that people would quickly become bored with just Looking At Webpages. This prediction turned out to be completely wrong. This is the problem with so many of the predictions about how much people will love 3D VR: in fact, we’re pretty happy with two-dimensional representations.

Else we’d have fallen with excited delight on 3D TV and 3D cinema. You may have noticed we haven’t. That’s why “the logical next step” is not the metaverse. It’s just Zoom or Teams and that sort of thing. Just as we’ve mostly stuck with books, not popup books. 2D TV and films, not the 3D version.
unique link to this extract

Using OpenAI in Rows • Rows

Rows is an online spreadsheet, and now you can incorporate OpenAI into it as a sort-of smart function:


There’s many things you can do with ASK_OPENAI in Rows:

• Clean Up Data: Capitalize text, remove unnecessary text, parse email domains, and more
• Text Classification: Tag emails, classify emails or classify roles from LinkedIn profiles
• Sentiment Analysis: This is a subset of Text Classification to identify and extract subjective information from text. It can be used to determine the overall sentiment of a piece of text, as well as to identify specific opinions and emotions within the text
• Translation: Translation into other languages
• Summarization: Condense news articles, messages, legal documents, research papers, and more
• Create Tables with data: Create tables with dummy or public data
• Find and Enrich data: Generate dummy data or public data points for things you already have on your spreadsheet, like countries and companies.


It’s hard to figure out if this is trivial or enormously helpful; Google used to (still does?) have a system for cleaning up data, but this can do more than that.
unique link to this extract

Boris Johnson erased from Grant Shapps spaceport picture • BBC News


Business Secretary Grant Shapps has denied doctoring a picture on his Twitter feed to remove former Prime Minister Boris Johnson.

Mr Shapps posted a picture of a ministerial visit to Spaceport Cornwall, as he hailed the first rocket launch from UK soil.

But social media users spotted that Mr Johnson, who also went on the trip in 2021, had disappeared from the picture. Mr Shapps says he was not aware it had been edited and has now deleted it.

An identical [almost identical – Ed.] photo remains on the Number 10 Flickr account, dated 9 June 2021. It shows Mr Johnson and Mr Shapps side-by-side in Cornwall during a visit to LauncherOne.
The then-Conservative leader is wearing a jacket emblazoned with the words: “Prime Minister”.

A source close to Mr Shapps [ie Shapps himself – Ed.] said: “Grant wasn’t aware anyone had edited the picture. He removed it as soon as it was pointed out. Obviously he wouldn’t endorse anyone rewriting history by removing the former PM from a picture. He was proud to serve in Boris Johnson’s government.”


Suspect it wasn’t Shapps, but an over-eager spad (special adviser) tasked with putting out something encouraging showing Shapps and the rocket but who thought, oh dear, can’t have Johnson there – it would show it’s old and also BJ is persona non grata.
unique link to this extract

• Why do social networks drive us a little mad?
• Why does angry content seem to dominate what we see?
• How much of a role do algorithms play in affecting what we see and do online?
• What can we do about it?
• Did Facebook have any inkling of what was coming in Myanmar in 2016?

Read Social Warming, my latest book, and find answers – and more.

Errata, corrigenda and ai no corrida: none notified

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.