If you’re into cooking and into physics 🙋🏻♂️ you CANNOT miss this paper on how to prepare the perfect pasta cacio and pepe! 🧀🍝
While the equations in the middle might scare you off, skip to the perfect 2-person recipe! 🧑🍳👨🏻🍳
“For a practical example, consider preparing Cacio and pepe for two hungry people. This typically requires 240 grams of pasta (tonnarelli is preferred, though spaghetti or rigatoni also works well) and 160 grams of cheese. Traditionalists would insist on using only pecorino, but some argue that up to 30% Parmesan is acceptable, though this remains a point of debate. To achieve the correct starch ratio, 4 grams of starch is optimal for 160 grams of cheese.”
Read the rest below 👇🏻
And, if you’re Italian, leave a comment on whether you think having Parmesan cheese is acceptable!
Reading Simon Willison‘s excellent blog on “Things we learned about LLMs in 2024”, made me realize why we’re not seeing much economic benefit* from #llms yet.
Simon writes:
“Most users are thrown in at the deep end. The default LLM chat UI is like taking brand new computer users, dropping them into a Linux terminal and expecting them to figure it all out.”
What should we work on, then?
As it was always the case with AI, work on the pains and the gains of the end users and think about a UI and UX they can use and be more productive with!
Don’t expect them to change how they work simply because you turned on a feature on your Office or GitHub subscription!
Quoting from The Economist: Artificial intelligence has already made many people—particularly shareholders in Al firms or chipmakers—very rich. But so far it has had little impact on the global economy.
One of the most amazing changes #genai has brought to the corporate world is that, while a talking head flips through slides, I don’t yawn at the trite corporate photo but I start wondering what insane prompt they used to generate the “art” on the deck.
Learn how we helped the largest supplier of spare parts for the agricultural and construction industries in Europe set up self-service analytics with dbt Labs, allowing Product, Sales, and Marketing teams to be onboarded in minutes, and Kramp’s analysts to access their data in a consistent and quick manner.
Why shouldn’t #data teams let the business query raw data and they are insisting on preparing the data?
My colleague Bo Lemmers shares her insights on the Xebia blog. The takeaways:
🐛 Misinterpreting the results is an issue. Do they know what units each column is using? Where do they find out? What if they assume and, as a result, mess up?
🔍 Missing records (or duplicated records). Without the data cleaning step, the business will get frustrated by handling these issues manually!
🐘 Slow results. When querying raw, unprocessed data, you risk scanning through billions of rows and columns that are irrelevant to your analysis!
💸 High costs. Raw data is not optimized for querying. No optimization, more money needed to query it!
⏳ Duplicate work. Every team will create their slightly different transformation, wasting hours and hours while work could have been centralized
🌦️ Misaligned decision making. If everyone is left to define key business metrics as monthly active users, there won’t be consensus around these metrics, leading to misaligned decisions.
It was a pleasure to be a guest on Tim Freestone’s Objective Hiring Podcast, talking about the Impact of #ai on Hiring.
Together, we’ve discussed how important it is—in this AI-infused time—to keep a human touch during your interview process, bias and discrimination, how to stand out, and more.
Recorded in the beautiful Xebia’s studio with a shout out at the end to - Luuk Feitsma—who taught me everything about recruitment and which I still miss every day!
