The Buzzfeed offices in New York were a microcosm of the company – but the tech industry only wanted to chew it up and spit it out, a former staffer says. CC-licensed photo by Anthony Quintano 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.
A selection of 11 links for you. Got a spare ribbon? I’m @charlesarthur on Twitter. On Mastodon: https://newsie.social/@charlesarthur. Observations and links welcome.
Unions representing Hollywood writers and actors seek limits on AI and chatbots • The New York Times
Noam Scheiber and John Koblin:
»
When the union representing Hollywood writers laid out its list of objectives for contract negotiations with studios this spring, it included familiar language on compensation, which the writers say has either stagnated or dropped amid an explosion of new shows.
But far down, the document added a distinctly 2023 twist. Under a section titled “Professional Standards and Protection in the Employment of Writers,” the union wrote that it aimed to “regulate use of material produced using artificial intelligence or similar technologies.”
To the mix of computer programmers, marketing copywriters, travel advisers, lawyers and comic illustrators suddenly alarmed by the rising prowess of generative AI, one can now add screenwriters.
“It is not out of the realm of possibility that before 2026, which is the next time we will negotiate with these companies, they might just go, ‘you know what, we’re good,’” said Mike Schur, the creator of “The Good Place” and co-creator of “Parks and Recreation.”
“We don’t need you,” he imagines hearing from the other side. “We have a bunch of A.I.s that are creating a bunch of entertainment that people are kind of OK with.”
In their attempts to push back, the writers have what a lot of other white-collar workers don’t: a labour union.
Mr. Schur, who serves on the bargaining committee of the Writers Guild of America as it seeks to avert a strike before its contract expires on Monday, said the union hopes to “draw a line in the sand right now and say, ‘Writers are human beings.’”
«
The point about the WGA (as it’s known) being an unusual beast, by being a union for white-collar workers, is very salient. Being pessimistic – or optimistic – about the potential for AI to evolve and develop is a sensible defensive position. Of course it isn’t close now. And it probably won’t be close to being able to write a scene for years. But you wouldn’t want your forebears to have sold your future for a mess of pottage, would you.
We interviewed the engineer Google fired for saying its AI had come to life • Futurism
Maggie Harrison spoke to Blake Lemoine, who Told You It Was Bad:
»
BL: In mid-2021 — before ChatGPT was an app — during that safety effort I mentioned, Bard was already in the works. It wasn’t called Bard then, but they were working on it, and they were trying to figure out whether or not it was safe to release it. They were on the verge of releasing something in the fall of 2022. So it would have come out right around the same time as ChatGPT, or right before it. Then, in part because of some of the safety concerns I raised, they deleted it.
So I don’t think they’re being pushed around by OpenAI. I think that’s just a media narrative. I think Google is going about doing things in what they believe is a safe and responsible manner, and OpenAI just happened to release something.
MH: So, as you say, Google could have released something a bit sooner, but you very specifically said maybe we should slow down, and they —
BL: They still have far more advanced technology that they haven’t made publicly available yet. Something that does more or less what Bard does could have been released over two years ago. They’ve had that technology for over two years. What they’ve spent the intervening two years doing is working on the safety of it — making sure that it doesn’t make things up too often, making sure that it doesn’t have racial or gender biases, or political biases, things like that. That’s what they spent those two years doing. But the basic existence of that technology is years old, at this point.
And in those two years, it wasn’t like they weren’t inventing other things. There are plenty of other systems that give Google’s AI more capabilities, more features, make it smarter. The most sophisticated system I ever got to play with was heavily multimodal — not just incorporating images, but incorporating sounds, giving it access to the Google Books API, giving it access to essentially every API backend that Google had, and allowing it to just gain an understanding of all of it.
That’s the one that I was like, “you know this thing, this thing’s awake.” And they haven’t let the public play with that one yet. But Bard is kind of a simplified version of that, so it still has a lot of the kind of liveliness of that model.
«
I still don’t believe Lemoine’s correct about the consciousness part, but the inside info about Google is fascinating.
unique link to this extract
Black holes resolve paradoxes by destroying quantum states • Science News
Lisa Grossman:
»
Don’t try to do a quantum experiment near a black hole — its mere presence ruins all quantum states in its vicinity, researchers say.
The finding comes from a thought experiment that pits the rules of quantum mechanics and black holes against each other, physicists reported April 17 at a meeting of the American Physical Society. Any quantum experiment done near a black hole could set up a paradox, the researchers find, in which the black hole reveals information about its interior — something physics says is forbidden. The way around the paradox, the team reports, is if the black hole simply destroys any quantum states that come close.
That destruction could have implications for future theories of quantum gravity. These sought-after theories aim to unite quantum mechanics, the set of rules governing subatomic particles, and general relativity, which describes how mass moves on cosmic scales.
“The idea is to use properties of the [theories] that you understand, which [are] quantum mechanics and gravity, to probe aspects of the fundamental theory,” which is quantum gravity, says theoretical physicist Gautam Satishchandran of Princeton University.
Here’s how Satishchandran, along with theoretical physicists Daine Danielson and Robert Wald, both of the University of Chicago, did just that.
«
This is a really quite puzzling – as in non-obvious, but logical – outcome, but it seems to ascribe a bizarre power to black holes that’s hard to square with something that’s just a big mass. The next, obvious, question is, well, how close exactly can you be to the black hole before it starts messing around with your quantum experiments? (And, presumably, your quantum computers on your gleaming new starship?)
unique link to this extract
How Buzzfeed News went bust • NY Mag
John Herrman:
»
Even back when I worked at Buzzfeed, it was clear enough that one of two things was likely to happen. Scenario one, which [Buzzfeed founder] Jonah [Peretti] embraced and preached, was a world where “social news” made sense, and running alongside the tech giants was the profitable and righteous way of the future. Publishers’ adjacency to social media wasn’t a temporary and inherently doomed state of affairs — it was bankable, and major investments in pre-profit digital media were rational. Scenario two was less appealing to contemplate. In this world, all ad-supported news — not just BuzzFeed — really was as fucked as it otherwise seemed to be when Google showed up, even before Facebook made its brazen bid to capture and monetize the online commons. From the vantage point of 2023, the history described in [Buzzfeed editor Ben Smith’s forthcoming book] Traffic sounds less like a story of entrepreneurial experimentation than an account of a recurring industry delusion. But it’s a delusion worth studying today as it threatens to manifest again. The tech industry will not ever save the media. It will sooner eat it alive.
There are many more books’ worth of material to write about the last ten years in online journalism, but I’d like to take a moment to emphasize the straightforwardness of the overall story. Just over a decade ago, a small group of social-media services became very large. Facebook, which had started as a place to keep up with friends, evolved into a tool for consuming media. This created a massive and sudden demand for fresh content, including, at the margins, news. Publishers old and new rushed to address the need, epitomized by BuzzFeed, which raised huge sums of VC money on the promise it could do so profitably, with maximally sharable and engaging content, some of which was sponsored. The newsroom portion of the proposition was straightforward, if incomplete. The platforms were hungry for stories, and what is a newsroom if not a machine for producing fresh and authoritative links, ready to share, comment on, or get mad about? And so this era, whatever it was, began.
What came next wasn’t much more complicated. Social media kept growing, ingesting and digesting the web around it, and sending some of its users back as readers in exchange. Its business model was an improvement, in nearly every way, on that of the news sites that were now supplying them with free content: bigger audiences, better targeting, and endless user-generated media. In the early days — let’s say 2011 to 2012 — there was a lot of windfall traffic for media companies. Random stories from years ago would suddenly have hundreds of thousands of readers, having been stripped of their original context and reshared by Facebook users. These new readers arrived in large numbers but didn’t really stick around. Their arrival was interpreted as an invitation. In hindsight, it was a warning.
«
So true. The tech industry doesn’t want to share. It wants to take. Everything.
unique link to this extract
UK readers may lose access to Wikipedia amid online safety bill requirements • The Guardian
Dan Milmo:
»
The Wikimedia Foundation, which hosts the Wikipedia site, has said it will not carry out age checks on users, which it fears will be required by the [online safety bill when it becomes an] act.
[Wikimedia UK chief executive Lucy] Crompton-Reid said some content on the site could trigger age verification measures under the terms of the bill.
“For example, educational text and images about sexuality could be misinterpreted as pornography,” she said.
She added: “The increased bureaucracy imposed by this bill will have the effect that only the really big players with significant compliance budgets will be able to operate in the UK market. This could have dire consequences on the information ecosystem here and is, in my view, quite the opposite of what the legislation originally set out to achieve.”
Rebecca MacKinnon, vice-president of global advocacy at the Wikimedia Foundation, has said carrying out age verification would “violate our commitment to collect minimal data about readers and contributors”.
The online safety bill requires commercial pornography sites to carry out age checks. It will also require sites such as Wikipedia to proactively prevent children from encountering pornographic material, with the bill in its current form referring to age verification as one of the possible tools for this. However, there is also a question mark over whether any of Wikipedia’s content would meet the definition of pornographic material in the bill.
«
This was presented on social media as OMG GUVMINT IS SHUTTING DOWN WIKIPEDIA. Except as the story here notes, there’s a questionmark – rather a big one, I’d suggest – over how you’d describe Wikipedia as pornography. It’s self-evidently an education and information site. The government’s description is that “all sites that publish pornography” will have to put in checks. You’d need to add a lot of pornography to Wikipedia to really make it fit that description, which would be a perverse way to prove that you don’t like that requirement of the OSB.
unique link to this extract
Language experience predicts music processing in a half-million speakers of 54 languages • Current Biology
Jingxuan Liu et al:
»
we used web-based citizen science to assess music perception skill on a global scale in 34,034 native speakers of 19 tonal languages (e.g., Mandarin, Yoruba). We compared their performance to 459,066 native speakers of other languages, including 6 pitch-accented (e.g., Japanese) and 29 non-tonal languages (e.g., Hungarian).
Whether or not participants had taken music lessons, native speakers of all 19 tonal languages had an improved ability to discriminate musical melodies on average, relative to speakers of non-tonal languages. But this improvement came with a trade-off: tonal language speakers were also worse at processing the musical beat.
The results, which held across native speakers of many diverse languages and were robust to geographic and demographic variation, demonstrate that linguistic experience shapes music perception, with implications for relations between music, language, and culture in the human mind.
«
They got people to respond at The Music Lab. The implication seems to be that tonal language speakers are less good at keeping rhythm. Don’t ask them to judge that scene in Whiplash, then. Rushing! Dragging! WHICH IS IT!
Prompt engineering techniques with Azure OpenAI • Microsoft Learn
»
This guide will walk you through some advanced techniques in prompt design and prompt engineering. If you’re new to prompt engineering, we recommend starting with our introduction to prompt engineering guide.
While the principles of prompt engineering can be generalized across many different model types, certain models expect a specialized prompt structure. For Azure OpenAI GPT models, there are currently two distinct APIs where prompt engineering comes into play:
•Chat Completion API
•Completion API.Each API requires input data to be formatted differently, which in turn impacts overall prompt design. The Chat Completion API supports the ChatGPT (preview) and GPT-4 (preview) models. These models are designed to take input formatted in a specific chat-like transcript stored inside an array of dictionaries.
«
If you need an introduction – and let’s face it, this is probably going to be in the sixth forum curriculum in a few years (or should be) – then this is as good a place as any to start.
unique link to this extract
Web3 funding continues to crater — drops 82% year to year • Crunchbase
Chris Metinko:
»
In the first quarter of the year, funding to VC-backed Web3 startups hit its lowest point since the very early days of the space as deal flow continues to slow.
Venture funding plummeted 82% year to year, dropping from $9.1bn in Q1 of 2022 to only $1.7bn, per Crunchbase data.
The funding number is also a 30% decline from the final quarter of last year, and the lowest total since the fourth quarter of 2020 — which saw only $1.1bn — when many people had never heard of Web3. [Many people still haven’t – Overspill Ed.]
Deal flow also continued its pronounced drop, as only 333 deals were completed in the first quarter — down from 369 in the previous quarter and a sharp drop from the more than 500 announced in Q1 2022. The total number of deals is the lowest since Q4 2020.
Perhaps nothing illustrates the differences between the first quarter of last year and the first quarter of the current one in terms of funding to Web3 startups more than the dramatic fall of big rounds.
In Q1 2022, VC-backed startups raised 29 rounds of more than $100m. That included massive raises of $400m or more by ConsenSys and Polygon Technology, as well as — of course — FTX and its US affiliate FTX US.
«
FTX? Gosh I wonder what happened to them. Bet all the VCs took a lot of guidance from them.
unique link to this extract
Elizabeth Holmes delays start of prison sentence with last-minute appeal • CNN Business
Jennifer Korn and Catherine Thorbecke:
»
Elizabeth Holmes won’t be starting her 11-year prison sentence just yet.
The disgraced Theranos founder was previously expected to report to prison on Thursday, but she will remain free a little longer while the court considers a last-minute appeal, according to a filing Tuesday night.
Holmes was sentenced last November after she was convicted months earlier on multiple charges of defrauding investors while running the now-defunct blood testing startup. Earlier this month, her request to remain free while she appeals her conviction was denied by a judge, setting her up to report to prison on April 27.
On Tuesday, however, Holmes’ legal team filed an appeal of the judge’s decision. As a result, Holmes can remain free on bail while the latest appeal is considered by the court, as per the court’s rules.
«
Gnnnnnngggh. However:
»
Holmes’ ex-boyfriend and former COO Ramesh “Sunny” Balwani was indicted alongside Holmes and convicted of fraud in a separate trial. Like Holmes, Balwani’s legal team delayed the start of his prison sentence by roughly a month with an appeal.
Balwani reported to prison last week to serve out his nearly 13-year sentence.
«
Oh well, the wheels of justice grind slow, but they do at least grind.
unique link to this extract
AI journalism is getting harder to tell from the old-fashioned, human-generated kind • The Guardian
Ian Tucker:
»
A couple of weeks ago I tweeted a call-out for freelance journalists to pitch me feature ideas for the science and technology section of the Observer’s New Review. Unsurprisingly, given headlines, fears and interest in LLM (large language model) chatbots such as ChatGPT, many of the suggestions that flooded in focused on artificial intelligence – including a pitch about how it is being employed to predict deforestation in the Amazon.
One submission however, from an engineering student who had posted a couple of articles on Medium, seemed to be riding the artificial intelligence wave with more chutzpah. He offered three feature ideas – pitches on innovative agriculture, data storage and the therapeutic potential of VR. While coherent, the pitches had a bland authority about them, repetitive paragraph structure, and featured upbeat endings, which if you’ve been toying with ChatGPT or reading about Google chatbot Bard’s latest mishaps, are hints of chatbot-generated content.
I showed them to a colleague. “They feel synthetic,” he said. Another described them as having the tone of a “life insurance policy document”. Were our suspicions correct? I decided to ask ChatGPT. The bot wasn’t so sure: “The texts could have been written by a human, as they demonstrate a high level of domain knowledge and expertise, and do not contain any obvious errors or inconsistencies,” it responded.
…If the chatbot were a bit more intelligent it might have suggested that I put the suspect content through OpenAI’s text classifier. When I did, two of the pitches were rated “possibly” AI generated. Of the two Medium blog posts with the student’s name on, one was rated “possibly” and the other “likely”.
I decided to email him and ask him if his pitches were written by a chatbot. His response was honest: “I must confess that you are correct in your assumption that my writing was indeed generated with the assistance of AI technology.”
But he was unashamed: “My goal is to leverage the power of AI to produce high-quality content that meets the needs of my clients and readers. I believe that by combining the best of both worlds – human creativity and AI technology – we can achieve great things.” Even this email, according to OpenAI’s detector, was “likely” AI generated.
«
Requiem for the newsroom • The New York Times
Maureen Dowd:
»
“What would a newspaper movie look like today?” wondered my New York Times colleague Jim Rutenberg. “A bunch of individuals at their apartments, surrounded by sad houseplants, using Slack?”
Mike Isikoff, an investigative reporter at Yahoo who worked with me at The Washington Star back in the ’70s, agreed: “Newsrooms were a crackling gaggle of gossip, jokes, anxiety and oddball hilarious characters. Now we sit at home alone staring at our computers. What a drag.”
As my friend Mark Leibovich, a writer at The Atlantic, noted: “I can’t think of a profession that relies more on osmosis, and just being around other people, than journalism. There’s a reason they made all those newspaper movies, ‘All the President’s Men,’ ‘Spotlight,’ ‘The Paper.’
“There’s a reason people get tours of newsrooms. You don’t want a tour of your local H&R Block office.”Now, Leibovich said, he does most meetings from home. “At the end of a Zoom call, nobody says, ‘Hey, do you want to get a drink?’ There’s just a click at the end of the meetings. Nothing dribbles out afterward, and you can really learn things from the little meetings after the meetings.”
When Leibovich got his first newspaper job answering phones and sorting mail at The Boston Phoenix, he soon learned that “the best journalism school is overhearing journalists doing their jobs.”
Isikoff still recalls how excited he was when he heard his seatmate at The Star, Robert Pear, the late, great reporter who later worked at The Times, track down the fugitive financier Robert Vesco in Cuba. “Hello, Mr. Vesco,” Pear said in his whispery voice. “This is Robert Pear of The Washington Star.”
With journalists swarming around Washington for the annual White House Correspondents’ Dinner and cascade of parties, it seems like a good time to write the final obituary for the American newspaper newsroom.
«
I haven’t been inside a newsroom for a long time, but they did seem to be getting quieter. The biggest trend was away from boozy lunches and towards sandwiches at a desk.
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: Apologies: I forgot to include a link to a news story from 1995. Tomorrow we’ll have two years to cover!
The Overspill: when there's more that I want to say 