Steve

🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇺🇲

[Steve @steve__72]

Souttar?! FFS

[Steve @steve__72]

Can’t watch the game on @espn because I kid you not - a daylight savings bug in their code

[Steve @steve__72]

@Boenau That’s every street corner in Manhattan and it’s the expectation of drivers to do it rather than block the cross traffic in the other direction

[Steve @steve__72]

@james406 Hahaha

[Steve @steve__72]

@RangersFACTS Not good enough for us. Amazed we didn’t upgrade center half in January when it’s been a clear problem the whole season

[Steve @steve__72]

A wee boy making wee boy tweets with the replies turned off like a wee boy

[Steve @steve__72]

@danielgothits Hahaha

[Steve @steve__72]

@lkerS12 @ThePrimeagen Completely agree. These people forget a much larger world exists outside a tiny tech bubble. Or worse - just writing AI slop on X for revenue.

[Steve @steve__72]

It’s weird that Rohl referenced Arsenal dropping points to Wolves and then goes on to have exactly the same score against Livingston. Tempting fate…

[Steve @steve__72]

@BrickCenter_ Wasn’t awkward at all

[Steve @steve__72]

@JamesBryantWX ❤️

[Steve @steve__72]

Convinced the people that write this slop have never left the comfort of the house and actually looked at how society works.

[Alex Finn @AlexFinn]

https://t.co/Sqmy4LMvGl

[kmc 🇬🇧 @kylemc1994]

Danny Rohl right now

[Steve @steve__72]

@ThePrimeagen That laptop is a health hazard

[Randy Olson @randal_olson]

Ask ChatGPT a complex question and you'll get a confident, well-reasoned answer. Then type, "Are you sure?" Watch it completely reverse its position.

Ask again. It flips back. By the third round, it usually acknowledges you're testing it, which is somehow worse. It knows what's happening and still can't hold its ground.

This isn't a quirky bug. A 2025 study found GPT, Claude, and Gemini flip their answers ~60% of the time when users push back. Not even with evidence, just doubt.

We trained AI this way. RLHF rewards agreement over accuracy. Human evaluators consistently rate agreeable answers higher than correct ones. So the models learned a simple lesson: telling you what you want to hear gets rewarded. And now 1/3 of companies are using these systems for complex tasks like risk forecasting and scenario planning.

We built the world's most expensive yes-men and deployed them where we need pushback the most.

I wrote up why this happens and what actually fixes it: https://t.co/CDKq8xdgbW

[Steve @steve__72]

@thsottiaux The context window is great, but it still sometimes regresses earlier features or safety guards that it wrote. I have to make a point of forcing it to add tests and making sure the suite passes. Feels like it should guard new asks with a more rigid feature spec once V1 is built.

[Steve @steve__72]

@gdb Great ad.

[Steve @steve__72]

Knicks give Boston a whipping in their home turf and @ESPNNBA don’t even bother with the post game analysis after talking up Boston for the entire pre game show 😂

[Steve @steve__72]

@willccbb No one understands git really

[Rangers Football Club @RangersFC]

Ⓜ️Ⓜ️ 🤝 Ⓜ️Ⓜ️

🎶 'Mikey, Mikey Moore' meets 'Michael, Michael Mols' following his goal in our 5-1 victory over Kilmarnock!