What does performance mean? Well, it depends on who you are. The PE firm and the owners of the company always win. They never lose money regardless of what happens to the company. The employees in the short term always lose because they will either be let go or operating expenses will be "optimized" in a way that doesn't favor employees. The customers generally lose because the products and prices will be "optimized" where optimization refers to company finances and not product quality or prices. In the rare case where the PE firm actually improves and grows the company, the employees and customers might also win, but those are the employees and customers in the long term.

Apple dominates the top-10 list, and yet the top-10 list only accounts for 19% of all sales. What that shows is that there are a lot of different companies selling phones.

To show that Apple "controls the premium end of the mobile market," better statistics would be gross margin or percentage of global profit.

Will Google indemnify and make whole if the AI agent either makes a mistake or does something that is not desired? Having AI gather information is one thing even if the information is not entirely accurate, but agentic actions that involve money or commitments is something else. Or will Google have some mechanism that allows that user to verify the actions before they're done?

"cancer and Alzheimer's disease are rarely found in the same person"

"cancer diagnosis was associated with an 11% decreased incidence of Alzheimer's disease"

If cancer and Alzheimer's are rarely coincident, then why is there merely an 11% decrease in Alzheimer's given that cancer exists? I would expect the decrease to be close to 100% unless "rarely" is actually a fuzzy term that is far from 0%.

Yes, the current Trump administration has canceled many science-based programs, people, and ideas. The really big question is how permanent the Trump influence will be after 2028. My take is that most Republicans currently in office are carpetbaggers rather than ideological extremists. They see the opportunity to advance their own careers by riding on Trump's coattails. Those coattails likely disappear after 2028, unless another Trump 2.0 person arises and tries to repeat the Trump strategies.

Another factor in restrictive zoning is discrimination against other races and lower social economic classes. It's not an accident that low density, single family homes are mostly owned by rich white people. In these neighborhoods, apartments are negatively correlated with school test scores, which significantly affect home values.

Yet an other factor in the US is the huge disparity in the quality of neighborhood schools. This desire toward "better" schools creates an artificial scarcity even in the face of a huge abundance of land.

There's a difference between a technology being used as a tool for innovation and that technology producing a packaged tool or report with a finished innovation. As an example, digging up and testing different microbes in randomly collected soil is a valid technique for finding new antibiotics and microbes with useful characteristics. In no way does this technique point to a specific innovation, but its use leads to specific innovations. Likewise, LLMs can lead to new innovations with having to produce a neatly finished, packaged answer to queries.

It's hard for us to think about the spinning mechanical disk TV because it's so different. A light shines on a spinning disk that has holes in a spiral pattern such that the light rasterizes across an object. The light bounces off the object and is detected by a sensor which then transmits the state of the sensor to the viewing device. The viewing device has the same type of spinning disk in front of a light that turns on or off based on the transmitted signal.

There are several problems with this technology. Only one person can view the image, and that person must be exactly lined up. Eventually Baird created a projector system. The rasterization is limited by the rotational speed of the spinning disk, so the images are not that clear, and the resolution was limited. A bright light to illuminate the object was needed, and that projected a lot of heat into the object, enough to make it uncomfortable for a human. The spinning disk also required a big, loud motor.

The whole point of the fake photos is to get foot traffic. Potential buyers who never walk in the door won't even consider buying the home, so the fake photo moves the buyer to at least consider buying the home.

This is especially important in the areas of California where supply is extremely constrained.

Fake photos are a form of bait and switch, and unfortunately bait and switch does lead to more sales.

The big difference between China and India is the ability of their governments to implement economic policy and large-scale projects. The Chinese have a dictatorship/autocracy which can implement policy forcefully and immediately without political challenges. And the Chinese have a money-printing machine with their export-driven economy. Meanwhile, although Modi and his party may dominate the national government, the Indian government system is unable to develop large-scale projects like power grids and large-scale solar, wind, or hydro generation, at least not on a scale even close to what the Chinese have done. The challenges at the government level include corruption and the lack of true dictatorial powers for Modi. India also faces a lack of funding, both at the national level and at the consumer level, and that funding problem will likely not improve due to the governmental problem. These two challenges are not going to change in our lifetimes.

Another way of looking at AI is that some students will use AI chatbots as a tutor to stimulate their own thinking and learning. Smart students use all the resources they have to obtain new ideas, generate questions, and then augment their personal understanding. These resources could be teachers, textbooks, classmates, etc. Now AI chatbots can be added to the set of resources.

Then there are students that will copy and paste (or slightly modify) AI output for assignments and projects, bypassing personal learning in order to make their lives easier and potentially raise their grades. However, this type of behavior is a reflection of personal characteristics. These students existed in the pre-AI age, and they similarly faked or plodded their way through school without mastering concepts. Pre-AI resources included Cliff Notes, smarter buddies, cheating strategies, etc. The availability of AI doesn't dumb down these students because their internal personality determines their ability to learning independently of AI.

Perhaps one could argue that the temptation of AI making life easier changes the former into the latter. But I would argue that the personality of the former largely avoids that trap.

Since forever and way before the pandemic, truly top talent has been able to work from home or wherever they want. But that's only for truly top talent where the guy would be hired even if there's no position. Most people who think they're top talent actually aren't because the company can find other people who are equally competent.

It looks like the Indian court order affects DNS. There is a blurb in the article about global DNS registrars being affected, with the implication that the ban extends beyond India. However, DNS is already implemented by country and region, so there's no technological reason to extend the Indian court order beyond India.

From a legal standpoint, no country wants another country to dictate what internet traffic is legal or illegal. Imagine having China, Russia, and the US issuing global takedown orders. We would quickly have almost nothing accessible remaining online.

Anthropic might have a specific idea of what safe, ethical, compliant, and helpful mean, but I seriously doubt it. These words are aspirational terms that have almost no utility in software specifications. Even in the looser world of vernacular speech, those terms vary across individuals and situations. If humans have differing definitions and interpretations of those terms, how can computer systems designed by humans be constrained by those terms?

Of course, this assumes that human software designers know how to constrain computer systems based on those terms. If Anthropic knows how to do this, they should patent it because it would be a magical innovation.

My guess is that the Anthropic constitution's main reason for existence is for sales and marketing.

The huge, huge caveat is that the NHTSA claim is based on self-reported Tesla numbers and is perhaps questionable and misleading. Many articles claim that Tesla accidents as reported to NHTSA are heavily redacted and trimmed. Tesla's definition of an "accident" is different and much more narrow than what NHTSA and other car makers generally use, so many crashes that almost all people would call an accident are not accidents to Tesla.

Furthermore, independent reports show that Tesla drivers are the worst drivers of all car makes and that Tesla cars have the highest fatality rate of all car makes. These reports support the notion that Tesla fudges its numbers.

Of course, all of these numbers likely characterize Tesla drivers rather than Tesla cars. It likely says nothing about Autopilot or FSD. It would be interesting to compare the accident rate for Tesla driven by a human versus the computer.