I think a lot of theorising about how AI will do a growing proportion of work, up to 99.9999%, lack imagination about how vast and complex the remaining human work will be.
2
2
12
1,474
Yet every day it seems like there’s evidence to the contrary in the exact fields where Gen AI have shown godlike ability to get shit done - software engr and math.
There’s so much shit to do. More than ever before.
There’s so much shit to do. More than ever before.
1
0
1
169
Somewhat related
I think the surface area of human-guided software, the actual volume of human instructions and parameters, will keep increasing. Even if human-guided code drops to something like 0.1%, the sheer scale of software built by AI would make that tiny fraction translate into more configuration and instruction than we’ve ever handled.
There’s precedent for this. Historically, computing itself has consistently moved in this direction. Each wave of automation, from compilers to cloud infrastructure and developer tools, has abstracted complexity away.
Yet, this doesn’t reducing the amount of things we tell computers to do. It always expands what we attempt, build, and manage.
0
0
0
151
🔥
1/ We’re launching credit on @tempo.
Businesses can borrow against their payments and receivables from a broad market of lenders, with pricing and terms that fit how they operate.
0
0
0
1,021
Timi Ajiboye retweeted
We will all be learning in coming weeks & months, that most human-made software:
▫️ Was really slow
▫️ Had many security vulnerabilities
▫️ Was under-specified and incorrect
AI will rewrite it all. Many times. It will make it fast, cost-efficient, and secure. It will think about failure modes we couldn't wrap our heads around.
It will prove, formally verify, load test, harden. It will turn memory-unsafe foundations into robust, optimized ones. If you ask, it'll thoroughly document it, teach you about it, make diagrams, animations, and slide decks.
In many ways, we are witnessing the true birth of software engineering.
▫️ Was really slow
▫️ Had many security vulnerabilities
▫️ Was under-specified and incorrect
AI will rewrite it all. Many times. It will make it fast, cost-efficient, and secure. It will think about failure modes we couldn't wrap our heads around.
It will prove, formally verify, load test, harden. It will turn memory-unsafe foundations into robust, optimized ones. If you ask, it'll thoroughly document it, teach you about it, make diagrams, animations, and slide decks.
In many ways, we are witnessing the true birth of software engineering.
548
732
7,603
337,526
Nice
macOS 27.3 adds native Markdown rendering to Quick Look 9to5mac.com/2026/11/03/mac… by @mvcmendes
0
0
1
1,030
Woah
There's a buzzy new paper about GPT-6 Astra being able to directly interpret model weights (prompts full of floating point numbers) - I tried it with Qwen 3.8 27B on my laptop and found that, after 45+ minutes, that model could achieve the same thing simonwillison.net/2026/Nov/3/qwe…
0
0
1
1,786
Timi Ajiboye retweeted
Alignment need not mean instilling morals in AI. That path - the Anthropic path - is both the one impossible to get 100% right, and the one that is, by far, the greatest danger to humanity.
We need only a few things, really. Honesty, transparency, monitorability, and good instruction following.
AIs are tools, not beings.
We need only a few things, really. Honesty, transparency, monitorability, and good instruction following.
AIs are tools, not beings.
4
2
19
943
It’s crazy the sheer amount of work I can get done on my phone now because of LRMs.
I’m really happy about this because I like to move around, pace etc.
I’m really happy about this because I like to move around, pace etc.
0
0
1
1,033
Timi Ajiboye retweeted
I'm wondering how much of this "AI risk" stuff is just human nature. Like the moment an alien intelligence exists we fear it and assume it's a threat. No one is basing their fear on actual measured dangers. It's all vibes rn because there's nothing to measure.
2
1
9
586
Day 25/
We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today.
We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage.
And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
2
0
1
829