OGWiseman Rounds Up!

Some interesting things I found on the Internet.

I can’t stop laughing about this post I found this week from the Dallas Fed, with its “projections” for A.I. in the coming years spelled out in all their glory. Specifically this chart:

This is a chart of GDP per capita in the future. The orange line is the long run trend. The green line (which you mostly can’t see because it’s buried behind the orange one) is the “what if A.I. is a normal technology” scenario where it boosts per capita GDP slightly for a decade or two.

The red line is “what if an A.I. singularity solves economic scarcity in the next decade?”

The purple line is “what if A.I. kills us all in the next decade?”

Thanks, Dallas Fed. It’s great to have some clarity, really proud of the work you did in this article!

*

From the “pretty sure it’s the red line” camp, here’s a podcast I found fascinating with the CEO of a robot company who has already started two *other* billion-dollar companies and so should probably be taken pretty seriously:

*

A couple of months ago I wrote this post, about the miracle effect that auto-driving cars are going to have on traffic fatalities, and ultimately human life expectancy.

Here are some relevant statistics. Just Tesla, just in their pilot stage program, has saved 75 lives, compared to the same number of human-driven miles. This is hardly a randomized study, but it’s real-world data over billions of miles traveled. Wow.

*

Story of a Beijing Vibe Coder. This is easily the most interesting thing I have read this month. A guy with a $200-a-month tier membership to Claude AI used $50,000 in compute in *one month* (running many parallel instances basically 24/7), and singlehandedly develops/maintains a dozen successful A.I. apps, to the tune of a million dollars a month of revenue with the inputs being his entire life and a $200 a month coding subscription. And honestly, even if Anthropic made him pay for the whole $50k of compute, a 95% profit margin is not so bad!

From there is springboards into a really interesting discussion of the Chinese tech scene, A.I. logistics problems, and the politics of the situation in both countries. Highly recommended.

*

From the other side of that ledger, the first A.I.-orchestrated cyberespionage campaign (that we know about) was just disrupted. To be clear, that doesn’t mean that A.I. “did it”, it means that a human automated A.I. to carry out the attack independently, and it did so. Coming soon to absolutely everywhere near you.

*

Sometimes the idea that we’re “building a mind” seems too abstract and underspecified as a metaphor for A.I. development. Then you see its brain, and it’s like, well, maybe it’s just early days. Gulp.

*

A.I. music is also getting better rapidly. If you listened without knowing, it would be hard to tell this wasn’t some Postmodern Jukebox ripoff. I don’t think “A.I. will destroy human art” is true, because you can still make all the art you want, I do every day, but I do think it’s going to make having an *artistic career* into something totally alien from what it’s been basically from the Caves of Lascaux until ~now.

*

I don’t know how real this is, but there’s now a serious ambition with money behind it to commercialize individual, door-to-door flying machines (“commercialize” = make it safe enough, cheap enough, and legal enough at the same time). They claim to have had some engineering breakthroughs and be making serious progress on the regulatory piece (which is frankly the hardest part).

I’m not always the earliest tech adopter, but if this happens, I will be in line bright and early. 8-year-old me would never truly have dreamed. Fingers crossed!

*

Speaking of first in line, Michael Mann’s long rumored sequel to Heat, the aptly names “Heat 2”, has been greenlit for 2026, with movie stars in every role. I feel positively tumescent at this news, and as far as I’m concerned, there’s absolutely no way that a sequel 20 years after the fact made by an aging director whose politics have become unseemly could be disappointing on any level. Hype train leaving the station!

*

This is extremely niche, but the guy who made one of my all-time favorite 90s computer games is now the head of Google’s Deepmind Lab, which helps make the Gemini series of A.I. models, Google’s flagship product.

In a full circle moment, the guy (Demis Hassabis) used Gemini 3 to vibe code a detailed version of his own 90s smash hit game. I don’t know who besides me needed this, but I presume there are dozens of us.

*

Oh yeah, I guess we cured blindness or whatever. Ho hum. Age of Miracles, etc.

END

Thanks as always for spending part of your Sunday with me! That’s what I’ve got for you on this one. I hope everyone has a great week, and I will be back next Sunday with something fun for you.