The 123rd Block: Liars and thieves
This week…
Malaysia’s Health Minister Khairy Jamaluddin publicly disclosed that his father-in-law, former prime minister Abdullah Ahmad Badawi has dementia at a conference on the management of dementia and schizophrenia. This was an open secret for years—his aides kept it well under wraps. I only knew about it because I worked in the media.
There were tell-tale signs, though. For example, he was the only high-profile politician who did not make a statement throughout the political turmoil the country has seen in recent years. He even published a book while his health was in decline, but his aides would only take email/written interview requests about it.
There is another high-profile Malaysian politician, from the Badawi generation, who is in a similar state. I have met him a few times at small, private, but social events. He would rely greatly on his wife to keep up the act, and everyone simply pretended nothing was amiss. One time, he pointed out my labret piercing and grinned. “Shiny,” he said. His daughter later told me it was the first time he had shown any recognition of anything in a while. It is poignant to see how cognitive impairment affects once powerful men so brutally, and the underlying shame it seems to bring to those close to them.
And now, a selection of top stories on my radar, a few personal recommendations, and the chart of the week.
Facebook parent Meta Platforms cuts Responsible Innovation team ($)
Jeff Horwitz for WSJ:
Meta Platforms Inc. has disbanded its Responsible Innovation team, which was once a prominent piece of its effort to address concerns about the potential downsides of its products.
The team had included roughly two dozen engineers, ethicists and others who collaborated with internal product teams and outside privacy specialists, academics and users to identify and address potential concerns about new products and alterations to Facebook and Instagram.
Meta spokesman Eric Porterfield said the company remains committed to the team’s goals, and that most of its former members would continue similar work elsewhere at Meta, though they aren’t guaranteed new jobs. He said the company believed its safe and ethical product design resources were better spent on more issue-specific teams.
A bit like when Google removed its “Don’t be evil” clause from its Code of Conduct?
Deepfakes are stealing the show on America’s Got Talent. Will they soon steal a lot more too?
Jeremy Kahn for Fortune:
Deepfakes are getting scarily good. If there were any doubt about it, this season’s America’s Got Talent should serve as a wakeup call. A startup called Metaphysic has managed to advance to the talent competition’s final round, which will air next week, by producing remarkable deepfakes of Simon Cowell and the other contest judges in real-time. The judges have been blown-away by seeing performers who have only the vaguest resemblance to them—a somewhat similar face and body shape—suddenly transform into their digital doppelgangers, right before their eyes.
Cowell reportedly said the deepfake act is “best of the season.”
Why does this horrifying woman keep appearing in AI-generated images?
Janus Rose for Vice:
The woman, whom the artist [Supercomposite] calls “Loab,” was first discovered as a result of a technique called “negative prompt weights,” in which a user tries to get the AI system to generate the opposite of whatever they type into the prompt. To put it simply, different terms can be “weighted” in the dataset to determine how likely they will be to appear in the results. But by assigning the prompt a negative weight, you essentially tell the AI system, “Generate what you think is the opposite of this prompt.”
In this case, using a negative-weight prompt on the word “Brando" generated the image of a logo featuring a city skyline and the words “DIGITA PNTICS.” When Supercomposite used the negative weights technique on the words in the logo, Loab appeared.
“Since Loab was discovered using negative prompt weights, her gestalt is made from a collection of traits that are equally far away from something,” Supercomposite wrote in a thread on Twitter. “But her combined traits are still a cohesive concept for the AI, and almost all descendent images contain a recognizable Loab.”
What I read, listen, and watch…
I’m reading an interactive piece by Manyun Zou, with Russell Goldenberg and Rob Smith, who watched 100 episodes of The Big Bang Theory on Youku, a Chinese video streaming website, to find out what the government in China censors, and why.
I’m listening to the audio version of Nature’s feature on hybrid brains on Nature Podcast.
I’m watching an episode on AI images on Last Week Tonight with John Oliver.
Reviews, opinion pieces and other stray links:
The focus on misinformation leads to a profound misunderstanding of why people believe and act on bad information by Daniel Williams on LSE Impact.
Inside a highly lucrative, ethically questionable essay-writing service by Chris Stokel-Walker for Input.
So, no more SIM card trays in iPhones. Is Apple getting too arrogant? by Navneet Alang for Toronto Star.
In between two lesbian worlds by Kira Deshler.
Chart of the week
From Know Your Memes, Aidan Walker writes an extensive insight into the top platforms where memes come from (2010-2022). In 2022, TikTok “overwhelmingly dominates meme origins” with 45.7 per cent of the share, followed by Twitter (30.1 per cent) and YouTube (8.6 per cent). Ten years ago, though, YouTube (34.9 per cent) and 4chan (28.2 per cent) were dominant, and trailing in third was Tumblr at 14 per cent. The graph below charts the relative shares of total meme origin for each platform over time. TikTok, although with a late start, is catching up to Twitter.