Dear Friend,
Apologies for the delayed delivery of this week’s Briefing – Jeffrey and I were both on the road all week and sometimes something has to give (a good reminder that humans are human after all). This week we’re introducing a new section, “Footnotes from the Future” – shorter morsels worth sharing that don’t warrant full commentary. Let us know if it’s useful.
And now, this...
Headlines from the Future
Students Are Catching Up With AI Job Changes: US computer and information science programs saw a steep drop in enrollment this fall – undergrad enrollment is down more than 12% and grad enrollment more than 10%. Over just two years, computer science programs have lost 18% of their undergrads and 27% of their grad students. It’s quite the reversal from the “become a coder and you’ll be set for life” mentality, which held until AI started eating entry-level coding jobs. Which raises the question: Which professions are next?
There Is No Moat in Frontier Models: Set aside the ever-shortening gap between US-based closed frontier AI models and open-weights Chinese models; US consumer spending further suggests that there isn’t a moat for the frontier model makers. A recent study analyzed 114 billion credit card transactions gathered from 174 million consumers and found that a mere 2.9% of those consumers pay for a major AI service – nearly all on basic plans of $30/month or less, and only 20% to 45% of them stay with a provider for a year – lower than many other subscriptions. They treat the services as close substitutes and switch whenever a better model ships. Without a moat around the models, companies like OpenAI and Anthropic need to build value-added services and lean heavily on B2B to justify their eye-watering valuations.
LLMs Are Coming for Intuitive (Creative) Tasks: A new study analyzed AI’s impact on the translation industry. Google Translate’s 2016 neural upgrade (predictive AI) cut demand for regular translation on a major freelance platform by about 9%, but had no measurable effect on “transcreation” – creative work like ad copy and poetry that needs cultural adaptation. ChatGPT (generative AI), on the other hand, cuts regular translation by 21% and transcreation by 30%. The “safe line” moved once, expect it to move again – and not just in language translation.
AI Agents Don’t Book Ubers: Turns out, just because you can doesn’t mean you should. Uber, in its earnings call, said that only four (yes, four) rides were booked by AI agents in Uber’s busiest week of October, Halloween – out of roughly 320 million trips and orders. It all makes sense when you consider that it takes you all of 50 seconds to order an Uber, but a bunch of back and forth with your AI agent. All of which is a good reminder that AI for the sake of AI doesn’t work – for users to adopt AI it needs to make things faster and/or better.
Footnotes from the Future
1️⃣ Science-Fiction’s whole deal is to predict a possible future – here’s how some of those predictions held up.
2️⃣ Apple’s App Store in numbers – answering questions like “What do most negative App Store reviews complain about?”
3️⃣ AI companies’ salaries are out of control – Anthropic’s IPO filing revealed that the average employee at the company makes $107,000 per month.
4️⃣ Meanwhile, in one survey, 19% of US workers said they’d been replaced by AI and later asked back.
5️⃣ European FinTech Revolut moved its coding AI to a self-hosted open-weights model and cut costs 90%. Expect this to become more common – and a big problem for OpenAI and Anthropic.
What We Are Reading
Silicon Valley Parents Want the Tech They Help Build out of Their Children’s Schools When the people who build the apps start pulling their own kids off them, it’s worth asking what they know, and this article shows how they’re quietly reshaping public schools. @Jane
After the Midterms, a Bigger Clash Over Data Centers Could Be Coming Anger over data centers is becoming a political force. With Democrats now in control of Congress, that could lead to new limits on where data centers can be built and how much say local communities have. @Mafe
The Jobs Question We’re going to need a more nuanced, human/social-systems approach to really anticipate how AI tools will change the many organizationally embedded jobs that don’t look like software development or accounting. @Jeffrey
If You’re in Tech in 2026, You’re Either at the Beginning of Your Career, or at the End of It (There Is No Other Option) I loved the DHH talk featured in this piece, especially his shift from thinking of himself as a programmer to a maker. I think many are having a similar identity crisis of sorts. There’s a bigger question here for anyone who’s questioning ‘who they are’ because ‘what they do’ is becoming obsolete: when the tools change what it takes to create something, how willing are we to become beginners again? @Kacee
In San Francisco’s A.I. Boom, a First Date Has Become an Interview Oh Silicon Valley… You just had to do it, I guess, and optimize truly every aspect of your life. Let the agents figure out that whole dating and love thing; meanwhile, we can philosophize about whether the machines become conscious or not. @Pascal
...and now for some fun!
🎨 I am a sucker for some fun, artsy, non-sensical thing running in a website – like this.
📺 Clip art was once everywhere. I remember selling CDs filled with them when I worked in a computer shop in the early 90s. And then – they just disappeared.
What We Do When We’re Not Writing
Hi! I’m Pascal from radical. When we’re not writing this newsletter, we help organizations turn volatility into advantage – without the “innovation theater.” We build your team’s capacity to handle disruption, so you stop reacting and start shaping. If that sounds useful, let’s talk.
↗ Dive into the deep end: Access our complete collection of 3,200+ radical links.