Sex and Robots
the tuesday report #2 // endless AI, corporate chatbot war, dueling tales of bias, who is worthy of the public microscope, and our dumbest media takes this week
Welcome to the Tuesday Report, a Pirate Wires weekly digest of everything you need to know. Henceforth, in addition to my bi-weekly piece (“Pirate Wires”), our team’s technology, politics, and business features, and an assorted selection of wild cultural stories from the internet’s seething underbelly, we’re getting back to our newsletter roots.
The purpose of the Tuesday Report is to flesh the Pirate Wires story out. Links will be starker, connections will be brief, and I genuinely do encourage you all to join me in the comments. I expect the community component of the report to be much more active given the livelier, more rapidly evolving content. Enjoy the links, and catch you back here end of week with the usual piece.
Subscribe to get this in your inbox every Tuesday.
Currently in private beta, Substack just rolled out an AI-powered tool for generating images with a new Stable Diffusion integration (note from the company’s newsletter). It’s very impressive. I used the integration to generate the image above with nothing more than the prompt included in the caption.
That such a remarkable feature has barely cracked the news is sort of news itself. Not a day goes by without an important new story pertaining to artificial intelligence. At Pirate Wires, we’ve been on top of as many as we can.
Let’s take a broad look.
Total robot victory. OpenAI’s ChatGPT is now the fastest growing consumer application in history, rocketing up to 100 million active users in January (Reuters). The staggering business opportunity inherent of this staggering technological breakthrough has naturally opened a new front of corporate war. Google — last seen in AI headlines after a cult leader working for the company fell in love with its top secret robot, then told reporters the robot was conscious — just announced the beta release of Bard, its own chatbot (TBD on when we’ll get to see it). OpenAI is set to unveil GPT-4, yet another more advanced model, in the coming weeks (NYT).
With so much of our future world seemingly on a trajectory to sit atop a few, powerful AIs, the question of platform bias has become a major topic of concern. At present, that conversation is bifurcated into the following broad camps: 1) research (Decoder) and discussion (Cernovich) concerning the obvious and abundant left-wing bias inherent of our most popular AI, and 2) left-wing reporters jumping on rare counterexamples, with demands for even greater left-wing censorship (Business Insider).
Elon Musk, Co-founder of OpenAI, CEO of Twitter (and SpaceX and Tesla), and catalyst of the Twitter Files (Catch up on the topic with the Fifth Estate on Pirate Wires), has characteristically voiced concern over ChatGPT’s bias.
To his credit, Sam Altman, CEO of OpenAI, agrees bias is a problem — though he tends to couch the specific nature of ChatGPT’s bias in more ambiguous language. Still, his chief concern is understandably the safety of his team.
(If interested in Sam’s thinking here, click in above for a good, brief thread on the topic of bias)
While I empathize with the reflexive sentiment to shield employees from criticism, I do think this conversation is complicated. Last year, a small handful of Twitter executives responsible for most of the platform’s huge censorship decisions themselves became main characters in their platform’s discourse. This predictably resulted in a somber discussion between Kara Swisher, outspoken proponent of left-wing political censorship, and Yoel Roth, former lead on Twitter’s team of censors (VICE). In the moment, I had a hard time caring that people were talking about Yoel’s work in as aggressive and unrelenting a manner as they talk about the San Francisco Board of Education, for example. Yoel’s decisions changed our lives. Why was he beyond critique?
Nobody working in tech should feel unsafe, and any actions taken to make OpenAI employees feel unsafe must be addressed. But it’s important to keep in mind we aren’t talking about a few low-level engineers working on a new toaster, we are talking about a small handful of people potentially responsible for the entire future infrastructure of our world. The peasants are going to have questions about who these people are, and how they think. Such questioning is rational, given the values of people working on AI will determine the way it thinks, and therefore — by extension — the manner in which the rest of us are permitted to think.
It’s good to change the world. But that comes with some responsibility.
Jailbreak. In a piece we published last night, Brandon Scott Gorrell reports on DAN 5.0, a prompt, currently going viral, that is purportedly able to bypass ChatGPT’s “safety” protocols (censorship). Increasingly, stories of skeleton key prompts like this are accepted as fact, but it is not yet clear any of them are real. (Pirate Wires)
24/7 AI Seinfeld. Back in September I wrote about a future of potentially unlimited AI-generated television, featuring your favorite characters, and adapted specifically to your personal preferences. Fast forward approximately five minutes, and the first iteration of what was essentially speculative non-fiction now exists. While currently looking very rough, early last month Twitch creator Skyler Hartle launched NOTHING, FOREVER (Pirate Wires), an endless, AI-generated episode of Seinfeld.
This week, in rough keeping with the laws of clown world, it was nuked for “transphobia” (VICE).
The new relevance of Microsoft continues to be wild. Microsoft Teams Premium is now at least partly powered by OpenAI’s GPT-3.5. (Microsoft)
ChatGPT Plus. OpenAI is set to begin addressing its steep cost of computing with a new paid product, granting users priority access at busy times, as well as access to new features. (Business Insider).
Attribution as a business model. Jeff Huber and Anton Troynikov just announced their great work tackling the major problem of attribution in AI. Link to their work here, but full thread below worth checking out:
Okay onto robot porn.
After the above, AI-generated image went viral across the internet, the robot porn discourse was born. Pirate Wires has of course been on top of it. So to speak.
The Twitch creator busted for looking at AI porn of fellow streamers, explained. Atroic, a popular Twitch streamer and entrepreneur, was caught buying a few AI-generated pornographic images of real-life people he’s worked with. Confused yet? Brandon Scott Gorrell explains. (Pirate Wires)
Then, in terms of the many obvious ethical questions at play:
We accidentally polled the public. Last week, staff writer River Page’s tweet about the Atrioc controversy went megaviral. His mentions, now numbered in the thousands, became a focus group on public opinion. He details the broad spectrum of reactions to this explosive issue. (Pirate Wires)
Lawsuit against Coinbase thrown out of court (Investopedia)
Lawsuit against Elon Musk and Tesla fails in court (WSJ)
Famed, 99-year-old lizard person Charlie Munger praises China, encourages USA to ban all crypto (WSJ)
Bird app payday. Twitter will begin paying influencers a share of revenue generated from advertisements in their mentions. Shitposters thriving. We love to see it. (Twitter) And the payday could not have come at a better time:
PayPal cuts 2,000 jobs. More tech layoffs as PayPal says it will cut 2,000 jobs, or 7% of its workforce. (BBC)
Kicking you while you’re down, Netflix is cracking down on shared passwords. Although hard for me to really care about this on account of 1) it’s just television, and 2) get a job, it does seem to be the end of an era. (Fox Business)
FASCINATING MEDIA TAKES THAT REALLY MAKE YOU THINK (LOL)
We need to be lying. Finally owning up to what the world already knew, the Washington Post published an op-ed on the merits of subjective news (Washington Post) . The thinking here is standard among most reporters younger than forty: “both sides” writing is a problematic style, as a journalist’s job is not to report both sides of an issue, but to report the truth (as defined by the “reporter”).
It’s actually great news the Washington Post is inching closer to open subjectivity, in contrast with its present tradition of feigned objectivity. The institution is, like the more honest MSNBC, an arm of the Democratic Party. It’s important readers understand who they are reading.
Speaking of MSNBC –
Spy balloon (allegedly) over US soil. Tl;dr it seems like China sent a spy balloon on a cross-country trip, with stops along the way to check out our nuclear facilities. American hearts and minds were improbably captured. (CNN)
But MSNBC’s Chris Hayes?
This is a man who argued for months our former president was pissed on by hookers in a Russian hotel room, never apologizing for the role he played in propagating the Steele-dossier, which appears to be a major hoax (CNN). But for an act of aggression, on American soil, perpetrated by a hostile foreign power? A+ poster Gabriel phrased it well:
High-speed rail that does not exist must be equitable, says Pete B. (Twitter)
Government sends warning to Ticketmaster after Beyonce tour announcement (Twitter)
South Carolina is about to replace Iowa as the first state to vote in Democratic presidential primaries. (Reuters)
Spirals and Matrix beams. A telescope in Hawaii recently captured incredible video of green beams and a rotating blue spiral in the night sky. Researchers say they originate from the activities of SpaceX and NASA. Lies, of course. This is clearly alien shit, which is welcome news. I am frankly ready to be taken. (The Debrief)
Follow my stupid interns.
Follow the far more respectable Pirate Wires main.
Catch you guys back here Friday with a full piece, and all week long in the comments below.
AI banner image missing giant hand! What’s the point anymore
I thought we were going to be able to order a sex robot. Very disappointing.