This week…
Your reading time is about 10 minutes. Let’s start.
Ps. On the back of Netflix and Amazon ceding to Indian authorities and removing anything unpatriotic, critical of Hinduism, challenging casteism, or otherwise offensive to the governing BJP party, I was going to write about the military-entertainment complex – what would be considered propaganda in many countries such as India, but somewhat normalised in Hollywood, but, maybe next issue, because…
Are you following the OpenAI soap opera? If yes, how closely? Because I tracked the updates as they came in; however, they moved strangely. It’s not exactly as complex as quantum physics to a non-physicist, but more like the labyrinthine plots and subplots of the reality TV shows my wife watches as she summarises them for me at the end. I try to keep track of which head of household has murdered a faithful(?) or which castaway won a shot in the dark(?), and which tops and bottoms have to lip sync for their lives... but it’s labouriously hard to keep up unless you take notes. So, I did. Well, not of who has sashayed away on this season of The Real Housewives, but of when is who doing what because of why in this OpenAI-Sam Altman saga.
So, here’s my box score for the exciting five-day test match held on the OpenAI ground, which ended in a draw:
Thursday, Nov. 16
All seems normal, but actually, Altman gets a message to join a meeting on Friday. OpenAI contacts CTO Mira Murati to tell her she is the next CEO.
Current board members: Sam Altman, Greg Brockman (OpenAI president, and chair of the board), Adam D’Angelo (Quora CEO and former Facebook CTO), Tasha McCauley (Center for the Governance of AI co-founder), Ilya Sutskever (OpenAI’s chief scientist), and Helen Toner (Georgetown’s Center for Security and Emerging Technologies director).
CEO: Altman
Friday, Nov. 17
OpenAI releases a statement that Altman is gone, and Brockman will step down as chair of the board. Brockman retains his role as OpenAI president.
Later that evening, Brockman announces he is quitting. A few senior OpenAI researchers also resign.
Board members:
Altman, Brockman,D’Angelo, McCauley, Toner, and Sutskever.CEO: Murati (interim)
Saturday, Nov. 18
In an internal memo, OpenAI COO Brad Lightcap says the board’s announcement surprises the management team.
Microsoft remains committed to the partnership with OpenAI, but other investors and some staff are angry.
The board members agree to resign and let Altman-Brockman back, but they miss the deadline.
Board members: D’Angelo, McCauley, Toner, and Sutskever.
CEO: Murati (interim)
Sunday, Nov. 19
Microsoft CEO Satya Nadella announces that Altman and Brockman have been hired to head a new advanced AI research unit.
Former OpenAI top talents are joining them at Microsoft, too.
Board members: D’Angelo, McCauley, Toner, and Sutskever.
CEO: Murati (interim)
Monday, Nov. 20
OpenAI announces a new, new interim CEO, Emmett Shear, the Twitch co-founder with shady tweets.1
More than 95 per cent of OpenAI employees sign an open letter to urge the board to resign or they will quit.
Sutskever also signs it because he “deeply regrets” participating in the coup.
Board members: D’Angelo, McCauley, Toner, and Sutskever.
CEO: Emmett Shear (interim)
Tuesday, Nov. 21
Business as usual on the user-end side as OpenAI releases ChatGPT with voice for unpaid users.
Microsoft offers to match OpenAI compensation to employees who want to join Altman-Brockman at Microsoft.
OpenAI CTO Murati is “still working on it” with Altman-Brockman, signalling whose team she’s probably on.
Late at night, OpenAI releases a statement of a “deal in principle” for Altman’s return and a new board with a new chair, who is a former Salesforce co-CEO. A former U.S. Secretary of the Treasury also joins the board. D’Angelo, who voted to remove Altman, stays. Brockman also returns.
Board members: Adam D’Angelo (Quora CEO and former Facebook CTO), Bret Taylor (former Salesforce co-CEO, and chair of the board), and Larry Summers (former U.S. Secretary of the Treasury).
CEO: Sam Altman, I guess.
So, let’s tally that up: Total CEOs (including interim) in the past week or so: three. Board members went from six to four, then to a net total of three – with two new faces, and five departures, including Altman and Brockman.
Three is such a tiny number for a board of such a big company, so they’re going to add more members, one of whom could be a Microsoft rep. Despite being OpenAI’s biggest funder, Microsoft never had a seat on the board. That’s probably partly due to the ‘hybrid’ governance at OpenAI, which is also likely why this whole fuss happened in the first place.
You see, when it was founded in 2015 by Altman, Elon Musk, and some other people (including Brockman and Sutskever), OpenAI was a non-profit entity. They simply wanted to develop AI tools to “benefit humanity as a whole, unconstrained by a need to generate financial return.” That’s grand, but three years later, Musk left. His departure meant losing access to his wealth, all while OpenAI came to terms with the expensive nature of its little venture.
So, the following year, OpenAI created a kinda-for-profit counterpart. It’s kind of like a for-profit thing, except it’s a ‘capped profit’ thing. That means investors can only make a certain amount if they make money. Anything beyond the capped amount is reinvested into the company. However, the non-profit board (and thus, its for-the-good-of-humanity value) continues to govern all operations. Hybrid model, yay!
Then, OpenAI launched ChatGPT to the masses last year. It was a Big Success, so of course it attracted Big Money from Big Investor. Although the board members (the Nov. 16 version) did not detail why they fired Altman, maybe they think he has been moving too fast by making too-powerful AI tools too commercial too soon. After all, they did try to replace him with a more AI-cautious Shear, who has publicly said that AI developments needed to be slowed down. So, hybrid governance, yes – but, sometimes, perhaps, maybe, a conflict at a fundamental level?
All that said, the board’s current composition looks more like a typical for-profit Silicon Valley board (of entrepreneurs, technologists, and politicians) compared to OpenAI’s earlier boards, which were comprised mainly of academics and researchers.
Anyway, this feels like a wacky episode of Legends of Tomorrow, and the loony legends stole the Time Bureau’s Waverider (again) with Gary’s help because he’s under Neron’s seductive influence (again), and they ended up messing up the timeline (again) and then had to do a quick and botchy fix before anyone noticed too much. If that doesn’t make much sense to you… Well, yeah.
And now, back to our regular programming. Here’s a selection of top stories on my radar, a few personal recommendations, and the chart of the week.
ICYMI: The Previous Block wagered that Osama bin Laden was the bigger story in last week’s news cycle than Sam Altman. CORRECTION NOTICE: Bad wager, this issue is my entire mea culpa.
Meet the Vietnamese grandmother fighting misinformation one YouTube video at a time
Bùi Như Mai and Lam Thuy Vo for The Markup:
Bùi had noticed a lot of right-wing misinformation being spread online by YouTubers in her community, particularly after Donald Trump was elected president in 2016. After she retired from her job as a software engineer at an aerospace company in 2018, she ramped up her efforts to fight back, broadcasting news on YouTube from her home in San Jose, California.
How to deepfake an election
Ondřej Kundra, translated by Alex Zucker for The Dial:
It appeared to be an audio clip of a conversation between Slovakia’s best-known investigative journalist, Monika Tódová, and prime ministerial candidate Michal Šimečka, whose campaign had promised voters that he would fight corruption, uphold the rule of law, and defend Slovakia’s integration in the European Union and western institutions. While Šimečka took a very critical stance toward Russia, his main opponent, the former prime minister Robert Fico, did the exact opposite. The conversation seemed to record Tódóva and Šimečka planning to rig the election in favor of Šimečka’s Progresivní Slovensko (Progressive Slovakia) Party by buying Romani votes, manipulating the vote count, and bribing journalists.
But the ‘conversation’ never occurred:
It was a deepfake, crudely cobbled together from other audio and video recordings and edited using artificial intelligence to produce a completely manufactured file.
In the recording, which lasts two minutes and thirteen seconds, there are several times when Tódová fails to complete a sentence, and when the intonation of her voice is off. She doesn’t respond right away to what Šimečka is saying — there are unnatural pauses in their discussion. But what Tódová told me that she saw as “half-baked” and Šimečka described as a “colossal stupidity,” others saw differently.
‘Your United States was normal’: has translation tech really made language learning redundant?
Ingrid Piller, Macquarie University for The Conversation:
With recent advances in automated translation, the belief is taking hold that humans, particularly English speakers, no longer need to learn other languages. Why bother with the effort when Google Translate and a host of other apps can do it for us?
In fact, some Anglophone universities are making precisely this argument to dismantle their language programs.
Even in the English language, there is great linguistic diversity:
The Australian meaning of “excellent” (from Aboriginal English) puts a spanner in the works. If you input “Deadly Awards” into any translation app, what you’ll get in your target language is the equivalent of “death-causing awards”.
What I read, listen, and watch…
I’m reading “Our Wives Under The Sea” by Julia Armfield. Creepy.
I’m listening to “A Tale of Two Tribal Nations” on NPR’s Code Switch, hosted by Anya Steinberg and Sequoia Carrillo.
I’m watching “How Dangerous Are Algorithms?” by DW Documentary.
Other curious links:
“‘Pallywood propaganda’: Pro-Israeli accounts online accuse Palestinians of staging their suffering” by Lara Bullens for France 24.
“The end of anonymity on Chinese social media” Caiwei Chen for Rest of World.
“How Musk’s X is failing to stem the surge of misinformation about Israel and Gaza”($) by Davey Alba, Denise Lu, Leon Yin and Eric Fan for Bloomberg.
“‘It’s not a public service, it’s toxic’: welcome to the world of gossip surveillance” by Alessandra Schade for The Guardian.
“Does Australia exist? Well, that depends on which search engine you ask…” by Josh Taylor for The Guardian.
“500 dólares al día por hacerte pasar por una modelo: así es el gran negocio tras los ‘chateadores’ de OnlyFans” por Marco Antonio Gomes en El País.
« Poutine dénonce un ‹ monopole dangereux › des Occidentaux sur les technologies de l’IA » par AFP dans Le Devoir.
Chart of the week
Martin Armstong illustrates how much global corporate investment in artificial intelligence has risen over the past decade. Read the report on Statista.
And one more thing
Jessica DeFino in The Unpublishable:
Between then and now, 11,000 Palestinians and 1,200 Israelis have been killed in the war in Gaza, and I’ve received dozens of PR pitches referencing the battle against skin fatigue, the battle against aging, battling dryness, battling dark under-eye circles, battling breakouts, battling acne-causing bacteria, going to battle against ingrown hairs and razor bumps, and battles with tangled strands. Still others encourage me to combat the 14 signs of aging, combat hyperpigmentation, combat oily skin, combat frizz to effectively eradicate repeat humidity offenders, combat future dryness of the hair and scalp, combat the long-term effects of winter on our skin, combat the dreaded dry skin, combat rosacea flare-ups, combat dark circles and puffiness, combat brassy tones, combat flyaways, combat adult acne, and combat all the variables wreaking havoc on my skin. Only an “army of products,” some say — a “skincare regime” — can help me “attack the dark spot process.”
Read on:
The Starting Block is a weekly collection of annotated media observations focusing on disinformation, data, and democracy beyond the Anglo-Western sphere. Read the archives. Find me on Instagram, LinkedIn, and Threads. Learn more about what I do here. Send questions, corrections, and suggestions to tinacarmillia@substack.com.
Al-Sibai, Noor. “OpenAI’s New CEO Posted Edgelord Tweets About Rape and Nazis.” Futurism, 20 Nov. 2023, futurism.com/the-byte/openai-ceo-edgelord-tweets. Accessed 25 Nov. 2023.
Fantastic read here. Well written and delivered!!!
Sorry for your bad wager 🤣 but jokes aside, great summary of the OpenAI drama and the references to several reality TV shows are well noted. 👍