The 1 individual human might learn from their horrible mistake, but not the other x billion humans (nor the many people every minute who drive for the first time).
And even the 1 human who made the mistake might forget it after some years, or make the same mistake again as they become elderly, etc.
Speeding and DUIs aren't mistakes, they're choices. There's a huge difference between the errors self-driving cars make and the equivalent errors humans make: the self-driving cars simply don't make those mistakes. In the category of things that self-driving cars do, insurance overwhelming handles those mistakes in both human and self-driving cases.
As far as I can tell (including looking at third party analytics attempts), there had been a massive increase in users over the last 3 years. Smaller communities tend to hold their trademark character a lot better. Pure speculation, but (beyond the bots) I suspect that a lot of the newer users are younger, and the attempt to be a bit more focused and sincere here is something they miss before they start posting.
You only just noticed this now? At the very least, HN is subject to the same intellectual capture that's taken over (seemingly) the whole damn world the past decade.
Seconded. Some common features: emotion-laden language; no new insights (let alone facts); low effort (poor punctuation etc). It's clearly a creeping problem here but I'm hopeful that the activist moderation and culling of politics-adjacent posts can keep a lid on it.
I’m always curious why local models aren’t being pushed more for certain types of data the person is handling. Data leakage to a 3rd party LLM is top on my list of concerns.
I am not as concerned with that with API usage as I am with the GUI tools.
Most of the day gig is structured extraction and agents, which the foundation LLMs are much better than any of the small models. (And I would not be able to provision necessary compute for large models given our throughput.)
I do have on the ToDo list though evaluating Textract vs the smaller OCR models (in the book I show using docling, their are others though, like the newer GLM-OCR). Our spend for that on AWS is large enough and they are small enough for me to be able to spin up resources sufficient to meet our demand.
Part of the reason the book goes through examples with AWS/Google (in additiona to OpenAI/Anthropic) is that I suspect many individuals will be stuck with the cloud provider that their org uses out of the box. So I wanted to have as wide of coverage as possible for those folks.
Worth noting that AWS Bedrock makes it easy to have zero retention with premier claude models. Not quite local, but it feels local-adjacent for security while getting affordable access to top-performing models... GCP appears to be a bit harder to set this up.
IMO Google Vertex is not any harder than AWS. AWS biggest pain is figuring out IAM roles for some of the services (batching and S3 Vectors -- I actually cut out Knowledge Bases in the book because it was too complicated and expensive). Have not personally had as big an issue figuring out Vertex.
I do have a follow up post planned on some reliability issues with the APIs I uncovered with compiling the book so much -- I would not use Google Maps grounding in production!
reply