Furthermore, it has nowhere near the impact of a $18 billion fine (counting all payments). 30% is predicated on YouTube and TikTok agree to similar payments and changes. So, it's (at least for right now), only 18 * 0.7 = $12.6 billion. Furthermore, that $12.6 billion will be paid out over 10 years. That's really $1.26 billion per year. Yes, that's something, but $1.26 billion compared to $18 billion for a company with $62 billion in post-tax annual profit makes a big difference.

How one labels others often reveals the spectrum location where one is situated.

"According to the Federal Reserve Bank of New York, the unemployment rate for recent graduates — which it defines as 22-to-27-year-olds with a new bachelor's degree or higher — was 5.7% as of June, more than the rate for all workers, which stands at 4.1%"

The unemployment rate may be 5.7% for recent graduates, but the underemployment rate is 42.0%. That counts people who are working at a job that doesn't require their college degree. The unemployment rate was as low as 3.9% in May 2022, when the underemployment rate was 40.7%. That means that the full-employment rate has been quite steady over the last few years. There has been very little change in job finding probabilities for a single individual.

However, there are factors that make a similar employment rate feel worse. College costs have risen, and today's graduates have a lot of debt, while inflation-adjusted salaries have lagged while inflation chugs ahead.

Is using AI to help edit worse than using Gramatik or a spell checker or a buddy?

"the FCC said the gigabit speed goal is not 'technologically neutral,' suggesting that it isn't fair to other, slower technologies."

DEI is evil when applied to people but is glorious when applied to network technologies.

Maybe it's in the terms of service. Comcast could monetized this data stream by selling to burglars who wish to increase their operational efficiency by identifying people on vacation. If this works out well, maybe Comcast could take a cut of the proceeds.

This is the one really interesting part of the study, or rather would be. However, even the abstract of the paper (which is paywalled) doesn't distinguish between sugar-deprived children who later increased sugar intake and those that didn't. Are the health effects still evident with sugar deprivation during childhood followed by higher sugar intake during adulthood? Or are the cardiometabolic disease, cancer, and telemere length effects mostly due to adult behavior, even if that adult behavior is influenced by childhood habits?

"loans to extend their monthly rent payment for a few weeks"

Instead of extending credit with an interest rate benefiting the creditor, how about having the borrower prepay the money and have the creditor pay an interest rate benefiting the borrower. We could call the creditor a "bank" and the scheme a "savings account."

A 99.98% accuracy rate will be wrong 1 out of 5,000 times. That's perhaps an acceptable Russian roulette probability for an individual, but for the store, that means hundreds (thousands?) of incorrectly identified people daily. If a person triggers the algorithm, what is the trained manager supposed to do? Is there a computer photo that the person can be compared against, or does the manager go with his Spidersense?

The biggest risk for students is an unmotivated student paired with unmotivated or incompetent teachers and resources. This has always been true, and nothing changes with AI. For motivated students, AI provides yet another resource that can be extremely useful, e.g., as a cheaper and more available tutor, but these students always found ways to learn even before AI or computers. For unmotivated students, AI could be used for cheating, but with or without cheating, the presence of AI for these students does nothing and changes nothing.

Perhaps AI might change the distribution of grades, as some cheaters will receive higher grades. However, cheating is as old as time, and grade redistribution due to cheating has likewise always existed.

This is just yet another way to inflate grades. Since the reputation of universities is not at all dependent on GPAs, there is zero risk for universities to inflate grades. Conversely, deflating grades leads to upset parents who send their donations and tuition dollars elsewhere.

This actually seems like a win-win situation. The students, parents, and university administrators and accountants are all happier. The future prospective employers are hopefully all smart enough to not base hiring decisions on fractions of GPA points, so it doesn't negatively affect the companies either.

So, in addition to badly trailing Nvidia in the AI GPU market, AMD is also aiming to badly trail Google, et al. in the ASIC space. The biggest customers for AI ASICs are the companies that design the ASICs themselves. That makes sense because those companies know exactly what they want the chips to do. It makes no sense for them to buy those chips from a third party because they already shift much of the supposedly non-core chip design to Broadcom, Marvell, or another company. ASICs only make sense in essentially embedded systems, so they make no sense to sell as a general, pre-designed processor. That's why GPUs are for.

Many companies prohibit employees from doing anything with their company stock other than holding the stock long. No shorting. No margin. No options. Why? One thing is to eliminate the incentive to manipulate their work for their own personal financial benefit. So, if Kalshi were serious about the stock market comparison, they would prohibit any employee, experimental subject, or others connected to the clinical trial from participating in Kalshi bets on that trial.

Or use streaming. One-to-one from the studio to Netflix. The studio gives Netflix the app with trivial authentication. Cheaper and easier than sending drives through the mail.

What does "drive better than humans" mean? Drive better in all possible scenarios? That's likely impossible now and probably forever. Maybe driver better in a majority of scenarios weighted by occurrence probability? Maybe, but we don't currently have the technology to validate this requirement.

Currently, we can cherry pick scenarios where some autonomous vehicles perform better or worse than the best drivers. What does that mean? Well, not much as a generalization for all AVs in all scenarios. Certainly any comparison to the worst drivers is useless. Comparisons to the "average" driver are also useless because the most average driver profiles are heavily skewed by the worst drivers. Can any AV currently perform better than the best human drivers? I would be shocked if that were the case.