Steven Sinofsky on why Jev fixes his oldest complaint about AI: natural language was never an efficient interface. "Ask yourself how many people are really, really good at asking questions. Less than half the people can ask a good question in a meeting." "And then how often do you... look at the answer and get really frustrated before it's finished, but you have to pay all this money to watch the seven paragraphs come out...?" "My favorite is the output of it is designed for probabilistic programming." "Instead of saying, 'Is this a customer service question? Then route to customer service, otherwise route to general help desk,' it's, 'This is 80% customer service.' That's exactly simulation." "There's 50 years of computer science research in literally probabilistic if statements. Suddenly the coolest place to be in computer science is gonna be probabilistic programming, which was all of computer science in the 1960s and '70s." "This is not all new. It's gonna be very interesting to dust off all of that work, because it's exactly what's going on. An if statement now is if X percent, not if always." "The way that Jev works is it's a custom programming language almost, which is: here's the prompt, come back with a percentage. And then you put that in the if statement." Steven Sinofskyshow more

ヨドバシカメラ😳

なんでヨドバシカメラのTシャツ着てんの？

ヨドバシTシャツにほっこり😊

Strong take. The durable version includes what would falsify it - not just what supports it.

The bottleneck isn’t just model intelligence. It’s making intent easier to express than a blank chat box.

if(true) getting replaced by if(mostly) is going to ruin me

Prompting was always a brittle interface.

The reason probabilistic programming stalled out in the 1970s was the cost of setting priors. Hand-crafting Bayesian belief networks meant engineers had to manually estimate every conditional probability table. LLMs didn't change the underlying control flow. They just gave us a fast, automated way to estimate those distribution priors on messy unstructured inputs. And once you treat them as prior generators rather than conversational buddies, software architectures get a lot cleaner.

So AI now decides pizza toppings, kidding

This is where AI gets more interesting for companies: less time reading its answers, more ability to use them in a workflow. But I’d want to know what that “80%” actually means. Has it been tested against real cases? What happens when it’s wrong? A percentage is only useful if you can trust it enough to decide what happens next.

we banned questions in meetings years ago efficiency improved immediately the meetings still take 4 hours

back to the future, literally. everything old is new again

half being able to ask a good question in a meeting is the observation that's actually quantifiable.

Probabilistic programming becoming the interface between agents and traditional software is a fascinating shift. It also changes the compute profile from long responses to many small decisions running in parallel.

🥳🤩

so isn't the real bottleneck teaching people to ask better questions vs making the interface better

Fuzzy Logic from the 80s.

Jev wins cuz choosing beats prompting

@grok can this statement about probabilistic programming in 1970s be verified?

None of my clients ask me anything. I fax them the renewal in October and they sign it. The insurance system works.