Dr. Joleen Liang is Co-founder of Squirrel AI, which pioneered adaptive learning at scale, first in China and now in the US. By 2021 its technology had served over 60,000 public schools in 1,200 cities across Asia. Squirrel AI has implemented large knowledge graphs mapping out the main concepts in the K-12 math, science, and language curriculum. The Squirrel learning tablet actively observes student behavior (including eye-tracking during video lessons) and adapts its presentation and testing to individual learning patterns.
this channel/podcast is excellent
very good analysis
"Today, we're breaking down the MIT study claiming 95% of generative AI pilots at companies are failing - and why this headline is misleading the entire market. The report, based on just 52 interviews and 150 survey responses, has been cited as a reason for AI stock crashes, but the methodology is deeply flawed and the findings are being wildly misinterpreted. What the study actually reveals is that while individual employees are getting massive value from AI tools (90% use LLMs regularly vs only 40% of companies buying subscriptions), organizations are struggling with implementation - not because the technology doesn't work, but because of leadership buy-in, poor change management, and organizational dysfunction."
Google Xâs former chief business officer Mo Gawdat says the notion AI will create jobs is â100% crap,â and even warns that âincompetent CEOsâ are on the chopping block. The tech guru predicts that AGI will be better at everything than most humansâechoing the likes of Google DeepMind CEO Demis Hassabis and OpenAI chief Sam Altman. Only the best workers in their fields will keep their jobs âfor a while,â and even âevilâ government leaders might be replaced by the robots.
Some of this sounds a lot like the hysteria about fracking a few years back.
I worked in a facility that used groundwater for cooling and I'm on a well myself. I'm curious what these facilities are doing with the water. The facility I worked in (a crystal growth plant) pumped water out of the ground, ran it thru heat exchangers, then pumped it back into the ground via an injection well. Inside the plant we recirculated DI water to cool the equipment—no net consumption. All we did we heat up the earth. I don't know why they would need to actually consume water unless they are just heating it up and dumping it into the sewage treatment system, which would also put a strain on that infrastructure, which the article doesn't mention.
If your well runs dry (been there) you don't notice it because only some of the taps in your house don't work. None of them work. They all feed from the same tank.
Like most articles intended to induce some kind of panic reaction this one is heavy on breathless anecdotes and light on technical details—likely because those reporting them don't have the background to understand them, but also because they don't talk to any of the targets of their wrath.
Wells (and plumbing in general) have all kinds of problems for all kinds of reasons. This case is not closed.
Every time a lawyer cites a fake case spit out by generative AI, an angel gets its wings. When the lawyers in Mata v. Avianca infamously earned a rebuke for citing an AI-imagined alternate history of the Montreal Convention, many of us assumed the high-profile embarrassment would mark the end of fake cases working their way into filings. Instead, new cases crop up with alarming frequency, ensnaring everyone from Trumpâs former fixer to Biglaw to â almost certainly â the DOJ. It seems no amount of public embarrassment can overcome laziness.
But so far, the system has stood up to these errors. Between opposing counsel and diligent judges, fake cases keep getting caught before they result in real mischief. That said, it was always only a matter of time before a poor litigant representing themselves fails to know enough to sniff out and flag Beavis v. Butthead and a busy or apathetic judge rubberstamps one sideâs proposed order without probing the cites for verification. Hallucinations are all fun and games until they work their way into the orders.
It finally happened with a trial judge issuing an order based off fake cases (flagged by Rob Freund). While the appellate court put a stop to the matter, the fact that it got this far should terrify everyone. (...)
He brings up a lot of interesting points at the end - things that may never be settled. Aside from his breaking down each track and finding those flaws, the random cliches and lack of continuity in the lyrics from that snippet reek of AI-generated text.
He brings up a lot of interesting points at the end - things that may never be settled. Aside from his breaking down each track and finding those flaws, the random collection of tired cliches and lack of continuity in the lyrics from that snippet reek of AI-generated text.
âThe rise of the internet brought about similar fears, yet it ultimately made learning richer and more accessible.â
Technology doesnât wait for policy, and as a current undergraduate student, I believe that the sooner schools catch up, the better we can use these tools to improve learning rather than undermine it. Still, an important question remains: Is it fair to compare AI with past innovations like calculators and the early internet, or is this a fundamentally different challenge?
AI is not the first technology to disrupt higher education. In the 1970s, the pocket calculator triggered a wave of backlash among educational institutions. Teachers warned that it would weaken studentsâ arithmetic skills, and some schools tried to ban calculators altogether. But others saw the potential: If students no longer had to do long division by hand, they could focus on bigger-picture math problems. Eventually, calculators became standard classroom tools, allowing students to shift their focus from manual computation to understanding formulas and solving higher-level, conceptual problems. Studies show that calculators can improve conceptual understanding when used correctly.
This same cycle repeated in the 1990s with personal computers and the early internet. Critics feared that spell-check and copy-paste would erode writing skills, and that search engines like Google and communal encyclopedias like Wikipedia would replace real research. And yes, some students misused those tools. But once schools embraced the technology and taught students how to use it well, evaluate sources, and cite correctly, their academic work improved. Students were no longer limited to the outdated books in their campus libraries, but suddenly had access to a multitude of books, articles, and datasets in multiple languages, at any time.
The cycle of resistance and delayed acceptance is a recurring phenomenon in large institutions, especially those with long-standing traditions in education, such as Columbia University. These universities, responsible for the education of millions of Americans, cannot afford to change course without serious caution. Even when faculty are eager to adapt, such as by updating policies on AI use in student essays, their efforts are often delayed by the universityâs complex bureaucracy and layered approval processes. These systems are designed to ensure thoughtful decision-making, but they can struggle to keep pace with rapid technological change. For example, a 2024 global survey conducted by the Digital Education Council found that 86% of students already use AI in their studies, underscoring the technologyâs rapid and widespread adoption across disciplines.
