Robert 🐀

I do things with the Internets and the Prints and stuff.

[Robert 🐀 @rkieru]

@SnazzyLabs Did you happen to notice that $50 difference is also getting you twice the storage (64 -> 128). It's not as though Apple were charging you $50 for an ethernet port...

[Robert 🐀 @rkieru]

@iabhi1610 @KirkMDesign

[Robert 🐀 @rkieru]

@pegobry_en Because it's moved from a novel thing to be solved, to a recognized factor in AI output. That doesn't mean it doesn't exist, or that people aren't mindful of it.

[Robert 🐀 @rkieru]

@bentlegen @rsms Wait… you use/built a tool but don’t know why it’s written the way it is?

[Robert 🐀 @rkieru]

@brankopetric00 Not sure I agree. They promised major investments in GitHub Projects years ago to make it a way to have your code and work management coexist. That never happened.

[Robert 🐀 @rkieru]

@thdxr Are you looking at whether that 15% investment returns its value (or greater)? Tying AI expense to a return on that investment seems to be a struggle as token costs increase.

[Robert 🐀 @rkieru]

@edandersen Who cares if AI could complete the task in a fraction of the time. If you enjoy coding; code. Recognize that AI does things fast- not synonymous with correct; secure; scalable.

Recognize that whole data centers are being built to approximate what your brain can already do.

[Robert 🐀 @rkieru]

@vlad_mihalcea Day one I kept it on "Auto" because I figured hey, I know the costs are gonna skyrocket why not take advantage of the 10% discount? Burned 17% in one day and it was garbage code.

Flipped from Auto -> GPT5 Mini. Using less credits and getting better results. Auto seems broken.

[Robert 🐀 @rkieru]

At this point GitHub's reliability stats have become a meme at work. We welcome 'Christmas' when automated alerts come in indicating a GitHub service is having issues (red) or operational (green)

[Robert 🐀 @rkieru]

@randomrecruiter A terrible but revealing answer. We're told to use AI, told to trust AI over years of experience. AI is going to 10x us, no, 100x us, no, another exponent! You don't need skill; just AI.

Move fast, don't think, trust AI. That's the plan.

[Robert 🐀 @rkieru]

@jour_n3y @RhoRider Where are you seeing this? From what I see, AI is becoming more expensive rather than lowering costs. Companies like MS are scaling back AI for that very reason, pivoting to more targeted uses than "Everyone spend as many tokens as you can! AI all the things!"

[Robert 🐀 @rkieru]

@sama Can it fix the massive disconnect between the goals of AI companies and the world economy? Close the inequality gap between the world’s richest and the rest of us? Cure cancer? Crack cold fusion? Solve colonizing the galaxy?

No? Just code and silly images? That’s… nice.

[Robert 🐀 @rkieru]

@intraalpha @wallstengine They could have invested in training, find new revenue streams and opportunities. By all accounts they had the time, money, and talent to take that route.

Instead they treated people like a liability, while people like you cheer them on singing ‘burn baby burn’

[Robert 🐀 @rkieru]

@DJ_CURFEW I wonder if you gave any thought to investing in improving the 22%. You want AI innovators? Grow them! What you’ve essentially done here is weaponize productivity. Beat your coworker for $$$. Come in last place and expect to be fired.

[Robert 🐀 @rkieru]

@Enelemtal @wallstengine What was improved? Is their business BETTER without those people? Based on their stocks it would seem that they HURT their business with this decision.

[Robert 🐀 @rkieru]

@intraalpha @wallstengine Coinbase had no financial need to fire them. They could have upskilled the staff, figured out how the staff could leverage AI to do more. If your AI strategy is to just replace humans with binary you're showing a lack of imagination. Use AI to help humans do MORE.

[Robert 🐀 @rkieru]

@Cold_horizons @crypt0lake AI is quickly becoming more expensive than humans, not less. That will eventually change but right now I think OP is right. Shedding quality for quantity.

[Robert 🐀 @rkieru]

@chelo_xl @hypeddev @reactjs I don’t understand? Am I missing something on MDN that identifies custom select as a web standard, or are you assuming that something existing on MDN makes it a web standard? Because MDN is a great resource, but not an authority on what is and is not a web standard.