A Trilli - AI, Green Energy, Uber, and Reading
AI
At this point, a significant portion of the tech industry runs on hype cycles - a burst of excitement over a new technology, which fuels a ton of investment as a bunch of VCs and founders rush to cash in. Then, depending on whether the tech is useful or any actual innovations occur, the cycle repeats itself a few times until it either becomes a mature, sustainable business (cloud computing, native advertising) with a few entrenched players, or it fizzles out (blockchain, metaverse).
AI is certainly nothing new, nor was it new in late 2022 when ChatGPT took the tech world by storm, but it had been in a hype lull, as most people used monikers like ML (machine learning) and Big Data. Then, OpenAI dropped a chatbot, and all hell broke loose.
The resulting AI hype cycle has gone on for over a year, and virtually every big (and small) company has embraced it fully. AI is everywhere, and it is going to do everything and solve every operational problem, because, well, AI. None of the CEOs hyping it can tell you what it does, exactly, but that's not stopping them. AI!
AI's head pitch man, Sam Altman, has also become the main character of both the hype cycle and the burgeoning ecosystem seeking to make AI as ubiquitous as cloud computing. The problem is, cloud computing was straightforward: what if we took a bunch of servers that people had to pay for individually, networked them all together, and used software to divvy them up into a bunch of (scalable) chunks for everyone to use?
This was an actual solve for a problem many companies faced - how much computing did they need? If you ran a retail business, the answer was 'as much as it takes to handle our busiest day of the year' but that wasn't a satisfying answer because for the rest of the year they'd be stuck paying for servers they weren't using. Cloud computing solved this for companies, entrepreneurs, and even individuals trying to do things like set up a personal webpage or run a blog.
AI, in most of its current iterations, is a solution in search of a problem. Those iterations also happen to be extremely energy and resource intensive, which means we're spending incredible amounts of time and money to do things without any clear benefit to society.
Despite this, Altman has announced he intends to raise between five and seven trillion dollars to fund the next generation of AI development. This is a preposterous number, so comically large I don't even need to offer comparisons to impress upon you its scale.
According to Altman, some would go to producing more of the expensive, energy intensive chips for AI computing. The head of NVIDIA, the company that makes the sophisticated chips Altman says we need more of, said this is ridiculous:
"You can't assume just that you will buy more computers," Huang said. "You have to also assume that the computers are going to become faster, and therefore the total amount that you need is not as much."
Another chunk of the trillions would go to building new data centers to house all the big AI chips but, again, Huang thinks the numbers don't add up:
Huang said: “There’s about a trillion dollars’ worth of installed base of data centers. Over the course of the next four or five years, we’ll have $2 trillion worth of data centers that will be powering software around the world.”
Here is where we need to pause for a minute to consider what both Huang and Altman are saying - if AI is left to grow unchecked, we will double the number of datacenters in the world within five years. Which is, I would argue, not moving in the right direction towards curbing energy consumption to stop ourselves from cooking the planet. The numbers are stark:
A recent report from the International Energy Agency (IEA) estimates that the global energy demand of data centers, AI, and crypto could more than double by 2026, increasing from 460 TWh in 2022 to up to 1,050 TWh — similar to the energy consumption of Japan. Meanwhile, in the United States, data center energy use could triple from 130 TWh in 2022 — about 2.5% of the country’s total — to 390 TWh by the end of the decade, accounting for a 7.5% share of total energy, according to Boston Consulting Group.
At current technological levels, it would be impossible to provide this amount of additional energy in anything approaching a 'green' way - unsurprisingly, Altman doesn't include trillions in his budget for building renewable energy to power these new datacenters, though he's 'invested in nuclear fusion' startups so maybe solving fusion in the next five years is part of his plan.
If we were to give AI the extreme benefit of the doubt, we could perhaps argue that the energy investment would be worth it, if AI could save us time and resources elsewhere? A hyper-intelligent AI could, like, cure cancer and save millions of lives and decades of medical research? Right?
...Alphabet chairman John Hennessy told Reuters that getting a response from Google’s chatbot would “likely cost 10 times more” than using its traditional search tools.
Oh. Okay. So Google is investing billions in adding AI to its search product, and it costs ten times as much to operate.
Even Sam Fucking Altman spends an alarming amount of time badmouthing his own chatbot:
"We're at this barely useful cellphone," Altman said this week at the World Government Summit, where he made a virtual appearance.
[...]
"We have not seen as much world-changing application as we'd like," Altman said.
What. Are. We. Doing. Here. The guy saying he needs $5 trilli also says the product he's built with a few tens of billi is 'barely useful' and not changing the world (for the better) at all. On that point, we agree.
Green Energy
In 2022, the US passed the Inflation Reduction Act, which dedicated billions to making American more green. It contained huge subsidies for developing and implementing clean energy solutions, with the stated goal of reaching 100% by 2035. With little more than a decade to achieve this feat, how's it going?
A nationwide analysis by USA TODAY shows local governments are banning green energy faster than they’re building it.
At least 15% of counties in the U.S. have effectively halted new utility-scale wind, solar, or both, USA TODAY found. These limits come through outright bans, moratoriums, construction impediments and other conditions that make green energy difficult to build.
Ahhh yes, of course. One problem with the way America does things is that even when the federal government writes laws and creates goals and guidances, local authorities can simply just...not do any of it. Which, listen, states and cities probably should have a decent amount of autonomy when, say, the Supreme Court decides to leave things (abortion, voting) entirely in their control while saying they can't do other things (administer elections, restrict guns). But the problem here is that our ambitious climate goals are being hamstrung by people who, for whatever reason, don't want to build solar and wind in places that are very sunny and windy:
“It’s 15% of the most highly productive areas to develop wind and solar,” he said. “Our overall goals are going to be difficult to achieve if the answer is ‘No’ in county after county.”
It's not necessarily where you'd expect, either. Sure, Tennessee has effectively banned development, but so have Connecticut and Vermont, not exactly reactionary enclaves. Texas has been scaling back its renewables projects for partisan reasons - cheap energy is Woke now.
But it's not simply the sunniest states that will save us - solar tech has come a long way since the 90's and both it and wind can be effective all over the country:
“A solar panel in my part of Michigan produces 70% of the solar power the same panel in Phoenix, Arizona, does. Most people would expect the difference to be much bigger,” said Sarah Mills, a professor of urban planning at the University of Michigan who studies energy policy and land use.
Which is why these local county bans are so counterproductive - by refusing to allow companies to build windmills and panel farms on empty land outside developed areas, politicians are creating a snowball effect for our climate goals.
Nor does it end there - fossil fuel companies are so opposed to the green revolution they are trying to influence nationwide building codes to make them less efficient:
The International Code Council, the nonprofit organization responsible for writing widely adopted model building codes, broke its own rules to allow natural gas trade associations make the industry’s case for scrapping provisions for electric appliances and car chargers from the latest update to the codebook, HuffPost has learned.
Long accused of inappropriately chummy ties with the industries its rules regulate, the ICC late last year abruptly changed its own written policies to give the gas groups twice as much time to file appeals against codes they don’t like, and to skip a key bureaucratic step meant to provide oversight to avoid frivolous challenges, according to public documents and interviews with four sources with direct knowledge of the process.
Which begs the question - are these same groups resisting at the local level? It certainly quacks like a duck:
In Ohio in December, an anonymously funded group held a catered town hall meeting in Knox County featuring speakers linked to fossil fuel and climate change denial organizations who made many unsupported claims. Representatives of the [solar] project were not allowed in.
And! Outside the fossil fuel-funded think tanks and astroturfers, the utility companies themselves are often incentivized to block green energy projects because they threaten their power monopolies.
It was good, and necessary, for the US to pass sweeping legislation and set aside large sums to greenify the country's power systems, but now the hard work of overcoming coordinated, well-funded local and systemic resistance begins.
Uber
Did you hear the news? Uber made a profit last year! A real one, not a made-up number if you ignore things like 'costs'. Good for them! How did they finally achieve profitability after almost a decade and a half?
Uber turned a net profit of nearly $1.9 billion in 2023, but what few of the headlines will tell you is that over $1.6 billion of it came from unrealized gains from its holdings in companies like Aurora and Didi. Basically, the value of those shares are up, so on paper it looks like Uber’s core business made a lot more money than it actually did.
Oh, right, the profits were from equity in other companies. Well, I suppose that's something. A look under the hood, however, reveals how Uber's ruthless cost-cutting and worker exploitation managed to just barely get them over the line into the black.
First, let's remember that Uber has lost over $30 billion dollars to get to this point. So even if they were 'for real' profitable to the tune of $2 billion a year, it'd take another fifteen years to get back to even.
But, leaving that aside, Uber's tactics are a rare case where a tech company is allowed to get big enough to use its monopoly power to climb out of the hole.
It did so in part by transitioning to 'upfront' pricing, a dynamic model that hides any sort of cost calculation from both driver and rider:
Dynamic pricing makes rider fares and driver pay much harder to predict by causing prices to shift according to more factors that are hidden from everyone but Uber itself, and it divorces the amount drivers are paid from the amount customers pay for the ride.
It allows Uber to control drivers by forcing them to take lower-paying rides to hit 'utilization' metrics or face consequences. Like the advertising auctions we talk about around here sometimes, creating opaque markets where the company in charge sets prices on both the buy and sell side means Uber can pay itself more with no repercussions - riders may grumble about fare increases, but they're unaware that money isn't ending up in the driver's pocket. The company openly admits to increasing the skim:
Before the pandemic, Uber was getting about 22% of the customer fare, but by mid-2023, that number had risen to 28-29%, amounting to a transfer of “over $1 billion in revenue per quarter from drivers to Uber shareholders.”
Taking a third more of each fare is definitely a way to make up for subsidizing rides for over a decade. It's a nice payday for Uber's CEO and leadership as well, who will now see their stock options increase in value (Uber's stock is up 10% on the news) and the company will use some of its newfound market cap to buy back $7 billion in stock.
Meanwhile, the company has fired a quarter of its staff, closed offices, and outsourced its expensive tech talent to India. For those behind the wheel, the situation is far worse - the company has successfully battled worker protections and is now cutting deals with taxi companies to further degrade the earning options for its beleaguered drivers.
Uber's business model was to take over a market - taxis - that didn't need and wouldn't benefit from tech optimization. And now that its forced taxis out of business and immiserated millions of people by trapping them in gamified gig slavery, it's declaring victory. At least on paper.
Reading
As a person who spends a few thousand words a week getting angry, it is nice to know a modest number of people (you) read them. Often, in these pages, we tackle complex topics that require (I hope) some level of critical thinking.
As a person way past college age, I am fully ignorant of what is happening with Kids These Days. It was therefore disconcerting to read this essay from a college professor, detailing what he sees as a marked decline in not only literacy but reading comprehension in students:
Even smart and motivated students struggle to do more with written texts than extract decontextualized take-aways. Considerable class time is taken up simply establishing what happened in a story or the basic steps of an argument—skills I used to be able to take for granted.
Nor is he alone - he cites multiple academic trade publications warning of a general decline in the sort of critical reading prior generations picked up during transit through the education system.
He is careful not to lay the blame on the students themselves, but faults instead the systems they are forced into. The No Child Left Behind Act sought to 'fix' what conservatives have long seen as a broken education system, not by better preparing and funding teachers but by imposing strict criteria on what success looked like in school - standardized testing.
The movement to impose nationwide standards at grade level began in the US in the 1990s, but it fully metastasized in the 2010s with Common Core. Teachers are now expected to 'teach to the test' more rigorously than ever before:
For all the flaws of the balanced literacy method, it was presumably implemented by people who thought it would help. It is hard to see a similar motivation in the growing trend toward assigning students only the kind of short passages that can be included in a standardized test.
Decontextualizing passages of writing because you don't have enough room in your curriculum to teach the entire book because you have to ensure a certain percentage of your students get a certain score on a standardized test or your school could lose funding and you could be fired sounds like...well, it doesn't sound much like what most teachers signed up for. And it is certainly not a recipe for students to emerge with the ability to process and, more importantly, enjoy reading.
Sure, the proliferation of devices and social media doesn't help. Students - and adults! - are barraged by dopamine feed scroll all day, and attention span and critical thinking do suffer.
What can we do? We've watched as the anti-education forces in this country gain more power, homeschooling or vouchering their kids while a once-great public school system strains under increased scrutiny. For religious and ideological zealots, the erosion of critical thinking perfectly suits the needs of their movements - armies of pliant drones to feed the machine. For society as a whole, we must hope we aren't losing an entire generation or two to dead-eyed YouTube explainers of the writing that molded our young minds.
Also, someone is going to have to write the newsletters once us old cranks run out of ideas.
Short Cons
ProPublica - "And so, the same state that questioned Mayron’s fitness to care for her four children forced her to continue a pregnancy that risked her life to have a fifth, one that would require more intensive care than any of the others."
NYT - "X, the social media platform owned by Elon Musk, is potentially violating U.S. sanctions by accepting payments for subscription accounts from terrorist organizations and other groups barred from doing business in the country, according to a new report."
The Independent - "“openai now generates about 100 billion words per day,” Mr Altman, who posts largely in lower case, wrote. “all people on earth generate about 100 trillion words per day.”"
Taylor & Francis - "We find that members of Congress share more links to low-quality sites than the public, that Republican members share considerably more than Democrats, and that this gap has increased over time."
Niedermeyer.io - "Tesla’s cynicism about its ostensible environmental mission was obvious from that first investigation, when I found them using diesel-powered Superchargers while claiming carbon reductions that assumed every electron that goes into every Tesla is zero emission."
CNBC - "Paramount Global is laying off hundreds of employees, just one day after the company announced CBS had record Super Bowl viewership, Chief Executive Officer Bob Bakish said Tuesday in an internal memo to employees."
Bloomberg - "“Eerily similar to WeWork’s public offering, we believe SHCO [Soho House] will eventually meet the same fate as the now defunct co-working space,” GlassHouse analysts wrote in the report.""
STAT News - "Ann Lewandowski, a health care policy and advocacy director at J&J, sued her company on Monday for allegedly overpaying its pharmacy benefit manager for its employees’ medicines, citing previous STAT reporting to support some of the allegations."
Know someone thinking of blocking local renewable power projects in their county? Send them HERE.