Aspirational Intelligence - AI, WinRed, and Click to Cancel
AI
There are two stories being told simultaneously about AI. One is that it is on the cusp of opening up a bright new future for humanity, eliminating broad swaths of boring work and freeing us up to live fuller, more meaningful lives. This version of the story justifies the hundreds of billions of dollars being poured into AI startups, billions more to build massive new datacenters, and the soaring stock values of any company even peripherally involved.
The other story is that AI is yet another tech bubble, a way for Silicon Valley to shine up its image following years of cascading scandals and lawsuits alleging that the tech industry has not been improving lives, instead concentrating vast wealth among a tiny, sociopathic elite. AI is the shiny object tech is dangling in front of our faces, hoping free chatbots keep us busy while they strip mine more of the economy for themselves.
Which story is true? In the past month OpenAI has declared it will shed its mission to work on behalf of society and become a for-profit company, handing its founder billions in equity. Though ‘for-profit’ is a stretch, considering the company is losing money on every ChatGPT query and continues to need billions in cash to keep the lights on. Meanwhile it continues to shed executives and other talent, many of whom are raising huge sums for their own AI startups, because why work for a fake non-profit run by a weird little tyrant when you can milk investors directly? This is not the behavior of employees at a company set to revolutionize the way the world operates.
Perhaps the cynicism is due to the fact that company’s ‘AI’ continues to struggle with things like not producing harmful gender and racial stereotypes based purely on a person’s name. It is yet another reminder than OpenAI is a complex set of computer algorithms trained on billions of pieces ‘scraped’ (stolen) data from the Internet, a place built and run by and for humans, many of whom have similar biases. Regardless of how much money OpenAI spends on building newer, faster models, or how much it pays low wage workers in other countries to flag problematic responses, as long as it continues to rely on Reddit users for new data, it’s going to sound like its mostly white, mostly male user base.
More worrying for the megafirms building datacenters for and investing billions in OpenAI and its peers are recent revelations that chatbots fail at basic comprehension:
Now, though, a new study from six Apple engineers shows that the mathematical "reasoning" displayed by advanced large language models can be extremely brittle and unreliable in the face of seemingly trivial changes to common benchmark problems.
Those ‘trivial’ changes? Adding irrelevant variables to a math question, such as describing the size of an item. If George has five large and five small kiwis, how many kiwis does he have? AI models failed miserably when presented with these minor deviations from the training sets they rely on.
These findings continue to emerge as researchers have more time and models to test with, because the major underlying flaw in the current iterations of AI is they are not performing reasoning, instead simply mimicking it based on complex pattern-matching based on vast data training sets.
Which, if the the tech founders behind AI were content to create more narrowly useful tools to do things like data analysis or research, would be really cool! If we called AI what it was - sophisticated pattern matching, whatever - instead of claiming the computers were becoming smarter than humans, we could credulously examine which tasks and jobs it could replace, or how it could make some of the least pleasant work tasks like writing emails or creating spreadsheets much easier. I have argued in these pages that these AI tools could (and maybe should) replace Google’s shambling ad golem with less biased search results. Maybe.
But you can’t raise money at a $150 billion dollar valuation with an email assistant company or ad-free search engine, so here we are. There is no evidence OpenAI or any of its peers are going to ‘surpass human intelligence’ or ‘fix the climate’ or any of the shit its hucksters insist is right around the corner (and a few trillion dollars.) The AI bubble looks like bubbles before it, and the people involved are not behaving as though they’ve discovered anything remarkable. The only thing life-changing is the sums of money they’re being paid to say so.
AI
Unfortunately for those of us forced to inhabit the Internet, many established tech giants are embracing it as well. LinkedIn is going to train LLMs on your banal work updates Google and Apple are shoving AI into our phones whether we like it or not. Microsoft has replaced Clippy (RIP) with a nearly impossible-to-remove ‘Copilot’ to apply its plagiarism algorithm to your Word documents and PowerPoint slides. Salesforce plans to cash in with an army of AI-generated ‘autonomous’ chatbots that will hopefully not promise customers anything companies can be held legally liable for.
Then there are the social media giants (and Twitter) who are all in on anything that lets them simulate user engagement. Meta has bet billions on an ‘open source’ AI model that will encourage wide adoption, because, unlike OpenAI, it has billions in ad profits to burn on vanity projects. What is Meta going to do on its own platform? That’s right, more AI slop for everyone!
And if chatting with Meta AI yourself sounds like too much work, the company will be testing a new feature on Facebook and Instagram that injects unsolicited AI-generated images “based on your interests or current trends” directly into your feeds, allowing you to “tap a suggested prompt to take that content in a new direction or swipe to imagine new content in real-time,” according to pre-event briefing materials.
[…]
Not to be outdone by OpenAI’s release of a voice assistant for ChatGPT earlier this week, Zuckerberg announced that users can begin talking to Meta AI and it will talk back in the voices of celebrities like Awkwafina, Dame Judi Dench, John Cena, Keegan Michael Key, and Kristen Bell.
Does this all sound deeply cursed and uncool? Mark Zuckerberg doesn’t care! His personal wealth depends on market analysts deciding that his slop-filled platforms aren’t hemorrhaging users, and every time you tap another Shrimp Jesus into your feed or talk to Judi Dench about the dog photo your aunt just posted, he becomes a few thousand dollars richer.
Not only is all this AI shit cringe to those of us who spend too much time Online, it actually turns consumers off too!
A study published in the Journal of Hospitality Marketing & Management in June found that describing a product as using AI lowers a customer’s intention to buy it. Researchers sampled participants across various age groups and showed them the same products – the only difference between them: one was described as “high tech” and the other as using AI, or artificial intelligence.
“We looked at vacuum cleaners, TVs, consumer services, health services,” said Dogan Gursoy, one of the study’s authors and the Taco Bell Distinguished Professor of hospitality business management at Washington State University, in an interview with CNN. “In every single case, the intention to buy or use the product or service was significantly lower whenever we mentioned AI in the product description.”
Let’s take a brief detour to ask: how did Dogan get this job? Why did Taco Bell decide to endow a professorship at Washington State University? I have questions.
Anyhow, the two main reasons consumers were repulsed by AI were lack of trust and transparency, after years of stories about Google search returning dangerous nonsense results and every AI company stealing user data to build their models. Huh! Maybe if AI companies had rolled their products out responsibly, the average person wouldn’t recoil at its mere mention.
AI somehow manages to be deeply flawed, stupendously wasteful, and widely unpopular, and yet the richest dipshits on the planet have decided we should invest the GDP of a first world nation on it every year until it…sucks less? What are we doing here?
WinRed
CNN has published an investigative piece on political campaigns preying on the elderly and dementia patients to reap millions in political donations. Despite the framing at the top that ‘Democrats and Republicans’ are doing this, let me direct you to this graphic further down the article:
If the name WinRed sounds familiar, congratulations you’ve been reading this newsletter for three years, because I wrote about these lawbreaking fucks back then.
For those unfamiliar, a little background. If you are operating a normal business, selling goods or services on the Internet, there are various rules about how you can advertise and sell. You aren’t supposed to make false claims, or use ‘pressure’ tactics like fake countdown clocks, et cetera. Many of these regulations are loosely enforced by the FTC, which is one reason scams proliferate in America more so than other nations.
But! There is another set of rules, the ones established and enforced by the card issuers. You may be able to set up your scammy storefront and evade the FTC’s notice for a time, but your merchant bank will have strict guidelines on how many refunds, customer complaints, or chargebacks you can have before they shut your account down. They will - if they’re doing their jobs - also regularly audit your payment pages and websites to ensure compliance.
Losing access to a merchant account costs companies not only money but is a major disruption to business. Therefore, even if their sales tactics may not be technically legal, merchants know better than to run afoul of the people who keep them paid.
WinRed, on the other hand, has used its political connections to carve out an exception to the rule. For years, it has run payment pages that would get any other company nuked from orbit - heavily abusing ‘pre-checked’ boxes opting users in to recurring donations. Here’s the donation page for a weird out-of-state fascist running for Senate where I live:
Note the bright yellow box with the checkbox ticked? Here is the donation page for my state legislator:
This ActBlue form requires an opt-in to activate a monthly donation. See the difference?
CNN puts a fine point on it:
The federal government has gone after non-political companies for similar tactics, such as making false statements in ads or making them seem as if they were written directly to the recipient. But regulators have done little to stop fundraisers from using misleading and deceptive advertisements to target vulnerable donors. And the lawmakers who experts say would need to act to protect consumers at both the state and federal levels are the same ones benefiting from the current fundraising machines.
Except, in this case, both parties are not benefitting equally. One party is using a private, for-profit fundraising platform violating the law by automatically opting people in to recurring donations, and the other is not! We just let them both do it, because political corruption seems to exist outside the purview of the legal system, and one party has ceded the argument to its more criminally-inclined rival.
Click to Cancel
Speaking of recurring subscriptions, the FTC recently released a rule requiring companies to make it easy for customers to cancel. So, of course, they are suing the government:
The NCTA - Internet and Television Association represents service providers like Comcast, Charter, and Cox and entertainment studios like Disney, AMC, Paramount, and Warner Bros. Discovery. The IAB’s 700 members include almost any company connected to advertising, with Google, Netflix, Amazon, Meta, Vizio, and the NFL among its board members, while the ESA covers home security giants like ADT.
The groups' lawyers argue that the FTC is trying to “regulate consumer contracts for all companies in all industries and across all sectors of the economy” by attempting to forbid businesses from making customers cancel services using a method that differs from how they signed up.
Yes, if there’s any group of companies I want defending my consumer rights it’s uhhh, cable companies, streaming networks, and home security firms. It is no surprise they are upset - no doubt many of them have spent dozens of hours building AI chatbots to make cancelling services even more difficult.
Unfortunately, in our current legal climate, deep pocketed industry trade groups can file endless lawsuits (in the Fifth Circuit, of course) to block common sense consumer protection rules any sane country would have enacted decades prior. For now, we can imagine a world where companies can be shamed into treating their customers with anything above contempt.
Short Cons
Bloomberg - “Traders at some of the world's biggest banks have allegedly been misled by cold callers dangling the prospect of jobs at the likes of Goldman Sachs Group Inc. and Morgan Stanley in exchange for details about their salaries, the make-up of their teams and even their desk’s confidential profit and loss statements.”
Balls and Strikes - “Three state Republican Attorneys General filed a complaint in federal court on October 11 arguing that their states have a right to pregnant teenagers and that right is being violated by the U.S. Food and Drug Administration.”
Bloomberg - “Elon Musk Is Now X’s Biggest Promoter of Anti-Immigrant Conspiracies.”
CBS News - “Under Trump's plans, Social Security's trust fund would become insolvent in 2031, which is three years earlier than currently projected by the Congressional Budget Office. At that point, the program would need to cut benefits by 33%, a steeper decrease than the 23% reduction forecast by the CBO in August.”
American Prospect - “The Republican governor of Montana couldn’t wear #37; Greg Gianforte is from San Diego. Its two GOP House members are Baltimore-born Matt Rosendale and Ryan Zinke, who is at least from Bozeman but who spends much of his time in Santa Barbara. And the hope of Republicans nationwide to win back the U.S. Senate rests on the shoulders of Tim Sheehy, a Minnesotan who got involved in Montana politics as a donor for other Republicans.”
The Conversation - “For instance, on roads where half of drivers are Black, Black drivers receive approximately 54% of automated camera citations. However, they make up about 70% of police stops.”
The Guardian - “Attacks targeting American public schools over LGBTQ+ rights and education about race and racism cost those schools an estimated $3.2bn in the 2023-24 school year, according to a new report by education professors from four major American universities.”
The Guardian - “When the University of Michigan governing board this year asked the state’s attorney general to bring charges against campus Gaza protesters, they tapped a political ally with whom some board members have extensive personal, financial or political connections, a Guardian investigation finds.”