I have no problems with the facts you are considering. But you did see the significant differences between changes in the telco supporting the I-phone and the two without it? Any retrospective survey is unlikely to ever prove a relationship, but this seems to be fairly strong to me in the absence of other evidence.

Complaints--
AI doesn't attribute prior research correctly
AI proofs are sloppy
Some mathematicians do not want to use AI
AI publication threatens the autonomy of academic mathematics.

This is all symptomatic with the publication process of the AI research, not the research itself. These academics (and I am one) are not making their case logically. No one wants to see professional jobs in their fields consumed by AI. However, it is unavoidably coming. These profs need to embrace AI if they are going to maintain their worth. As we all will have to do.

Sort of. The most successful publisher right now probably is the WSJ. This success (large circulation growth in a dying industry) is due to their new editor's focus on confrontation and sensationalism. She has taken a page out of AI companies and going all in on driving engagement. Not my cup of tea (she is British) but it seems to be working.

The price of this stuff will absolutely decrease, the only unknown is how companies will roll out new functionality and how specific future features fit into the pricing tiers over time.

I agree. The opportunity for AIs to act as Agents will be worth lots to providers. As always, we are the product.

Decline in reading by school age children is substantial, NAEP reading scores for older students have hit historic lows.

In 2024–2025 data releases:High school seniors (12th graders) had the lowest average reading scores since tracking began in 1992.
Only about 35% of 12th graders performed at or above "proficient" in reading, with around a third scoring below basic levels.

A problem from parenting, devices , and schools.

.Reading for pleasure has plummeted dramatically:Among 13-year-olds, the share reading for fun almost every day fell from 27% in 2012 to 14% in 2023 (nearly halved)

Traditional news sources (local tv stations and newspapers) are trying to reimagine themselves, their future audience has moved to screens, and podcasts have replaced long format journalism. With google's present focus on AI, dollars are going to other projects. California is in budget crunch and this is likely to get worse. Little chance that this will be successful, although many in government (both sides) would like to own a captive press.

The study awarded a point for each of five attributes. No real differences between 5 points and 4 points. Even at zero points, subjects were approximately 1 year older than the the best groups. Study looked to take ~ 10 years. Naive assumption (linear relationship) says over 30 years an additional 3 years of aging. Lots of things to worry about, but worrying your 90 year-old brain looks like it is 93 would not be at the top of my list. Of course, sleep is important, but this is not a doom and gloom result, imo.

Too many AI entrants; most will fail. Success depends on having a critical mass of users. So, specs-manship to attract users. BTW, other factors are required (for instance acquiring high-quality/novel training data) but the most important are number of users and engagement metrics.

Original: "We remember enough of the scientific literature of the times to know that you've never read a major science paper on the subject." Now --> "no peer reviewed models"? ..."By the end of the decade"? ... "and the second article you quoted" How about the first? Obviously there was scientific speculation, which was later judged to be unfounded. Otherwise, what pray tell ignited the media hype? Seems you are moving the goalposts? But a silly thing to argue, in the greater scheme of things.

I gently disagree. The research was on aerosols reflecting sunlight --> hence global cooling. There were definitely papers, just later found to be not correct. As far as "consensus" that was media hype --> nothing to do with reaching a real consensus at that point in time. The consensus against the theories came later, after more evidence was gathered. Some examples via Grok: 1971: Rasool & Schneider (Science) – suggested aerosols could cause cooling (later found to be overstated). 1975: Bryson and others – speculated about cooling from dust/aerosols. Several papers in Nature and Quaternary Research discussed Milankovitch cycles and the long-term likelihood of eventual glaciation (tens of thousands of years away).

Yes, but please take a sec and read the study. It is akin to saying I'm modeling a football game, where one player is randomly doing acts, the other has a single strategy no matter what the down and distance, and the other 20 players don't matter. It is not real world.

Algorithmic pricing can increase pricing. The most interesting/relevant current discussion on this topic is rental rates - I would guess you know This is not addressed by this study, and I still argue rarely seen in the real world. Requires only 2 players, each using one specific algorithm covered in the study, one of which is a random response to stimuli. In your experience, can you think of an example?

Yep, and highlights the danger I pointed out of a duopoly. You want to fix it, add other entrants. But this certainly does not apply to this algorithm, iows - still not real world.

I wouldn't get too excited about this. Almost nothing in the real world is a 2-player game. Is it extendable? What happens when other strategies are followed at the same time? Generalizing, it does imply the advantage of having a multiple of players, as opposed to a set duopoly. Being slashdot, lots of time the new entries are fueled by new technologies. In short, interesting, but not particularly applicable to the real world.