Not the fault of the Bluesky team - they are clearly following regulation, even if it is a nonsensible one considering nonresident travellers.

What a weird experience. And I don't like the idea of giving out IDs or personal information to third parties, at all...

The two are independent - but in the case of British Airways, neither has been built with anyone giving a damn clearly

It's an example where it will become less and less defensible when an AI agent can one shot better implementations (for the frontend, most definitely. BE is more nuanced)

If anything, AI will at least make it harder for objectively terrible software (in UX, performance etc) to exist (looking at you, British Airways app).

Because you can no longer say "it will take 6 eng months to fix it."

BS: it takes two prompts and 10 minutes! If you care

By games engines being commoditized: there are so many more games being built! And so it's harder to stand out unless you have differentiators (loyal fans, massive $$ for marketing, other clever stuff)

AI will result in so much more software that is average (or below) also!

I am instructing what the AI should write, correcting it, testing it etc

Simply it’s more efficient (faster) writing what I would write

When I used autocomplete to write the code I was still an engineer - this doesn’t change it.

And everting else - including accountability - stays!

Business Insider

The Register

Forbes

All have dramatic headlines that often don’t deliver but are clickbait

Almost all ad supported media - it is their interest

Ones that do not do it are subscription businesses: The Information, FT, Bloomberg (from what ai notice)

My longer analysis on what happens to software engineering, as a profession, if and when AI writes most of the code:

newsletter.pragmaticengineer.com/p/when-ai-wr...

... and this move also made S3 the perfect backend for large-scale, infinitely scalable databases like Turbopuffer and others!

This is the badge photo the “whistleblower” sent to Casey

Generated with AI ofc

Uber badges look nothing like this… there is no “UberEats” branded badge. AI isn’t good enough to know this, but later on it will probably improve and produce plausible looking badge photos!

We’re entering a time when it’s harder to trust anything online: and surely more people will try to fool journalists with AI-generated “evidence.” In some cases, they will succeed, especially at publications chasing headlines and not doing proper investigation / reporting

The Reddit thread: www.reddit.com/r/confession...

Casey’s interaction with the “whistleblower” where he gradually realizes all “evidence” is AI-generated, designed to fool even journalists… then he confronts the faker. Worth the read

www.platformer.news/fake-uber-ea...

Simon Willison told me something interesting: focusing on the theory actually holds back. These should be probability machines at best.

And yet, at scale, they are… unexpected.

I suggest reading The Bitter Lesson paper (very short)

Practice >> theory IME. Also recommend @simonwillison.net

For stuff that I can try out myself: I do it, and make my own judgement based on it.

Deep research is available to anyone - even in free plans (I pay $20/mo for a few vendors)

If you never try yourself, you likely stay a sceptic! Plus depend on others’ takes. Not how I like to operate!

I'm not talking about any promises though. And the expectation that devs will use AI tools for coding-related work in 2026 vs 2025 seems... like a fair assumption.

The whole post is about a CTO wanting to measure the "now" so they can compare if their expectation of more use manifests!

oh we agree we are not at the promised land... don't think we ever will be, with software!!

so you are suggesting don't measure anything?

The point is that this eng leader wants to get a baseline of how things are, today. E.g. what are defect rates, deployment frequency, devs' view of how their tools are etc. It will get messy once devs start to use more of these tools (which they will)

yes, the LLM collects more information, when using Deep Research. Details that were not in my original doc.

I'm ending this conversation now though. You keep asking: and I don't know why or to what end.

Try it yourself, or not. It helped me get a far better understanding of my case!

It's a CTO at a travel startup; VC-funded, profitable and growing, about 50 people AFAIK. Talked with this person today.

Here's someone who is not C-level, and not someone who buys into hype: Mitchell Hashimoto, creator of Terraform & Ghostty, now independent x.com/mitchellh/st...