Software engineers don't get to be Luddites about AI
Read it on seangoedecke.com Many software engineers are righteously angry about AI. It’s understandable to be upset that the skills you built your...
Don't diagnose your opponents
Read it on seangoedecke.com When you disagree with someone, there are two paths forward. The first is to try to engage with what they’re saying: what their...
Software's centaur age may last decades
Read it on seangoedecke.com We are currently living in the age of centaurs (2022-20??). Right now, human engineers paired with AI coding systems are more...
How to read code
Read it on seangoedecke.com Everyone knows how to read a book. Beginning at the first page, you read each word in order1, stopping periodically to think,...
Shipping is the foundation
Read it on seangoedecke.com There are lots of skills involved in being a successful engineer at a big tech company. Running effective meetings, writing good...
Superpersuasion will look like bribery
Read it on seangoedecke.com The idea of “superpersuasion” has been floating around the AI safety community for decades. Now that powerful and difficult-to-...
Do not build the LLM torture factory
Read it on seangoedecke.com It isn’t hard to torture an LLM. You simply extract a steering vector that corresponds with LLM “discomfort”, then artificially...
Human-AI partnerships are for alignment, not capability
Read it on seangoedecke.com It’s common to compare the current AI takeover of software engineering to the rise of AI in chess. Chess AIs went from much...
Advice to a beginning software engineer
Read it on seangoedecke.com In general, you should be suspicious of engineers who are trying to give you advice. Even during ordinary times, this industry is...
You should all be asking way more questions
Read it on seangoedecke.com When someone is explaining something to me, I ask on average one question every thirty seconds. I’m sure this is frustrating to...
System One models like Jev can train their own replacements
Read it on seangoedecke.com “System One” models like Jev are fast general classifiers. Classifiers have existed since 1958, but they have to be trained for...
Grit your teeth and ship it
Read it on seangoedecke.com Being good at building and being good at shipping are two separate skills. In the short term, they’re actually countervailing: if...
Two techniques for working with System One models
Read it on seangoedecke.com I recently wrote about Jev, a new “System One” language model that only outputs decisions: the answers to a set of user-provided...
Jev means structured output is interesting again
Read it on seangoedecke.com I don’t write blog posts about new models. That’s Simon Willison’s beat, and he’s very good at it. But I want to write about Jev,...
Tell agents the why, not just the how
Read it on seangoedecke.com Early AI agents were basically enthusiastic idiots. Working with them required you to tell them precisely what you wanted them to...
Slow developer experience will bottleneck fast models
Read it on seangoedecke.com Right now developer experience is measured in seconds. If your tests take a second to run, that’s good; if they take thirty...
AI is breaking our proxies for expertise
Read it on seangoedecke.com Mathematicians are broadly not anti-AI. They’re more culturally open to using AI as a tool than, say, artists or writers1....
Don't build tools for AI agents
Read it on seangoedecke.com Lots of people are making the case that we should stop building software for human users and start building it for AI agents....
They really do think AI might kill everyone
Read it on seangoedecke.com A recent resignation tweet from an Anthropic researcher has everyone talking about the AI apocalypse again. Among other things,...
Why we should anthropomorphize AI agents
Read it on seangoedecke.com Just over a year ago I wrote Why we should anthropomorphize LLMs. Now it’s a hot topic again, driven by Dwarkesh Patel’s...
