Mark Anderson

I live in NYC. I provide interesting links on #education. And stuff.

[Mark Anderson @mandercorn]

"The more Fox News a white American watches, the more likely they are to believe in a racist conspiracy theory, regardless of party affiliation and demographics"

https://t.co/QT5QJsTieZ

[Mayor Zohran Kwame Mamdani @NYCMayor]

New York City, we have no excuse to be bored this summer. The map at https://t.co/Wt1RgEfgey is full of free and affordable stuff to do -- sports, music, arts, movies, video games and more -- in every borough.

The city is your backyard. Get out there and enjoy it.

[Mark Anderson @mandercorn]

I've been having fun as a side project vibe-coding a jukebox on top of my digital music collection. An overview:

11,674 tracks, 90 countries, mostly lossless.

Roots reggae, ECM/fusion jazz, conscious hip-hop, and confessional singer-songwriters doing most of the heavy lifting.

Two-thirds instrumental. Sizzla and ECM Records lead the count.

[Mark Anderson @mandercorn]

And then making a visualizer that can make playing those playlists fun again!

[Linguistic Discovery @lingdiscovery]

A new database of features of languages of the Americas is now available!

[Nico Ajzenman @Nicolas_Ajz]

New WP 🥳

Can AI replace human counselors?

Large RCT (n=41K) comparing an AI chatbot vs. trained human counselors to motivate high school seniors toward teaching careers.

Summary: If anything, bots outperformed humans at a smaller cost and reduced human quality heterogeneity.

[Daniel Willingham @DTWillingham]

Intervention for at-risk readers (age 7-9) focusing on decoding and language comprehension shows positive results in RCT, including 6 month follow-up. https://t.co/PbvTvfwl30

[Mark Anderson @mandercorn]

To me this is in the "no duh" category, but apparently most of this country needs to understand this and here's another recent paper to back it:

wildly intensified immigration enforcement reduces the productivity of the immigrant population without providing any promised employment boost for U.S.-born citizens.

https://t.co/mbkMWVhEsH

[Scott Barry Kaufman ⛵ @sbkaufman]

"While AI assistance may ease initial learning, it appears to undermine the effortful processes needed for robust learning. These results have important implications of how generative AI tools should be integrated into higher education." https://t.co/Rsq9ugWvgc

[Mark Anderson @mandercorn]

"What we found is that there was very little difference for low-attaining young people between mixed-attainment and setting classes.

Meanwhile, for the high-attainers who make strong progress in high sets, they make lower progress in mixed-attainment classes overall.”

https://t.co/tqfB2qOMrq

[Mark Anderson @mandercorn]

@Nature As with humans?

[Mark Anderson @mandercorn]

I have large collections of digital music (lossless burnt from CDs) and writing (mine own) going back decades, so an interesting use case as a non-coder but Linux lite user (Ubuntu, Pop_OS) has been: what can I "vibe code" that would be most interesting to be able to explore and build upon from these collections?

[Mark Anderson @mandercorn]

I've been mainly using Gemini Pro for most things AI, but have decided to check out Claude Pro for a month. It's interesting to be almost immediately be able to see some differences. I'm not exactly a cutting edge user, so I figured you just pick one horse and run with it (as per @emollick https://t.co/Fs4aw6pM15)

[John A. List @Econ_4_Everyone]

My travels have convinced me that we are now in the Age of Emboldenment.

How so? First, the good news. AI is having an effect that is often not discussed: it gives the underconfident a voice. People who once struggled to articulate what they knew deep in their minds but had a difficult time finding the words can now communicate with much more clarity and precision. Giving the once voiceless a voice is wonderful.

But there's a dark side to this emboldening. AI is also giving people an awkward and truly borrowed confidence. Some act as if consuming an idea is the same as understanding it. With every new prompt, the AI answers come fast and furious, and with such seductive fluency. The sand castle looks magnificent. Then the questioning starts, the tide comes in, and the sand castle disappears.

The differentiator here? Using AI as an assistant versus using AI as a steroid. Using AI to help hone your own ideas and write them more clearly, organize your thinking, or stress-test your argument? That strikes me as a force multiplier. It makes genuine expertise more powerful and more visible, and this helps us all learn more from others' ideas. AI as an assistant.

But, relying on AI to generate ideas that you do not actually understand yourself? That is the steroid at work. And, I am not saying this as an academic who romanticizes/rewards struggle for its own sake. I see the same within organizations, even where total output, not struggles, is crucial. The premise here is that the struggle and wrestling with an idea are creating the understanding. You cannot borrow it. You cannot shortcut it.

Making real change within organizations almost always requires some sort of verbal discussion and defense of a proposed action, policy, or idea. The person who has genuinely wrestled with a hard problem and the person who asked AI to wrestle for them are not the same. Just like steroids get detected, fake knowledge gets exposed. Always has. AI just accelerates both the inflation and the reckoning.

So, let me end with a prediction that complements what I wrote earlier: critical thinking and communication will soon become, by far, the most valued skills in the room (maybe they always have been in some rooms, but even more important now). Homogeneity is the reason why: everyone has access to the same AI. This means that the differentiator isn't who can generate the answer, it is who understands and can explain it.

The Age of Emboldenment is here. Make sure your confidence is built on bedrock, not sand.

[Mark Anderson @mandercorn]

"Using an in-home HEPA purifier for one month spurs a small but significant improvement in brain function in adults age 40 and older."

https://t.co/5DjxFhP2sk

[Ev (like in 'evidence', not Eve) Fedorenko 🇺🇦 @ev_fedorenko]

Remember Chomsky's arguments about language innateness based on its putative un-learnability? Those arguments hinge on a crazy theory of grammar. Here is Ted Gibson's summary of the original claims and a proposed solution in the form of dependency grammar
https://t.co/nih32F53gk