Computer scientist. App & game developer. AI whisperer. 3D artist. Animator. Writer. CompTIA A+ & Network+ certified; Security+ in progress. I make things.
@ResilientCyber I'd like to see more real discussions on the topic. For example, what kind of security risks do vibe coders typically introduce into their products in practice? So far, I haven't gotten a single answer for this question.
@PeterBrochant@kennyistyping I actually read a few of the social science studies that were popular in the news. "Replicability crisis" is not a strong enough term to describe what's going on.
For me, the same day. If I'm not working at the bleeding edge of what's available to me and trying to disrupt an entire market, I'm not doing my job right.
Here's an interesting question. Will there come a time when your boss will fire you for not following an AI's recommendation? The moment that happens, the AI becomes your boss.
@slopmath Sure, I understand that AI isn't ready to completely take over yet and that wasn't my point. Human judgement is necessary, but that's on a continuum not an on-off switch. People are ALREADY taking "orders" from AI at least some of the time. We just don't call it that.
To be fair, even people at the top will be following "orders" from AI. If it is consistently giving you the best solution, what is the difference between an order and a suggestion? Assuming you want to be successful and competitive, it's a questionable distinction.
@mattpocockuk Why do you need to build a ubiquitous language? What does that even mean in practice? Can't you refactor from any language? AI is already quite good at porting between them.
It's possible to find huge efficiency gains in token use. People are able to do so much with AI that they often don't think about how they could do it more cheaply.
None of this is satire.
β Jensen Huang says a $500,000 engineer should burn at least $250,000 a year in AI tokens
β Meta staff burned 73.7 trillion tokens in a month on an internal leaderboard. Some left agents running idle just to climb it
β Meta is now putting teams on token budgets
β Apple reportedly cut engineers' daily AI limit from $300 to $200, with $25 next for some teams
β Uber capped AI coding spend at $1,500 per engineer per month, after burning its 2026 budget by April
β Goldman took Claude Code off its bankers' desks
β Revolut moved 2,000 engineers to a self-hosted model and says the bill dropped 90%
β A law firm told our team their Claude bill went up 10x when their plan switched to usage pricing
Burning tokens doesn't equal delivering value.
Anything you've learned can usually be relearned. I won't say coding is like riding a bicycle, but it'll probably come back faster than you fear if you ever need it again. Learning to program the first time literally rewired your brain in ways that make it easier to do again.
JUST IN: Vibecoder discovers previously unmapped ancient Maya settlement hidden beneath the Mexican jungle using 48 AI agents & NASA lidar data, despite having no archaeology experience.
I have to say. Making a Blender addon makes me appreciate it that much more as an app. Once you start having to consider all the things that they've solved, you really see how impressive it is.