Hacker News new | past | comments | ask | show | jobs | submit | timmy-turner's comments login

They will sell "training data slots". So that when I'm looking for a butter cookie recipe, ChatGPT says I'll have to use 100g of "Brand (TM) Butter" instead of just "Butter".

Ask it how to deploy an app to the cloud and it will insist you need to deploy it to Azure.

These ads would be easily visible though. You can probably sell far more malicious things.


They do but we have for instance education to reduce their hallucinations in narrow fields of expertise. And a system of guardrails to only let educated people work in those fields to avoid harm.

The above person was comparing it to friends and random online comments, though. I wouldn't be surprised if AI is far more reliable than those.

No, Dresden was a big garrison city, manufacturing and transportation hub. Bombing it, as cruel at is was, served a purpose for the Allies to remove safe retreats for the Germans.

Lots of Dresdeners were supporters of faschism before the war and after the war. A quite vocal minority still supports neofaschism up to this day. In the 90ies they installed the myth of the "holocaust bombing" as an allied war crime, trying to turn themselves into the victims of war.


> Lots of Dresdeners were supporters of faschism before the war and after the war.

Loads of Afghans support the Taliban. Doesn't mean firebombing any city there is the proper answer to anything there.

And loads of people support the US government, the Russian government, the British government, and so on and so on. All of those countries have some absolutely massive atrocities under their belts.

And absolutely none of it is deserving of burning the flesh off civilians and children.


Another quite common headline technique is "Car hits pedestrian/bicycle/other-car". As if the car was driving itself. It somehow shifts the blame away from the driver to the object, the car. Similar to accepting that software has bugs and thus anything can happen an no one is to blame.

It sucks to read this, but its completely sound and logical for those who only drive cars. As more people start using bicycles these days, its slowing shifting though.


To realize how skewed this really is you just have to imagine a world where "a bicycle hit a pedestrian". You will never read that, it will always be "cyclist" with the implication of "your fault for even using a bike"


I’m not sure I agree with you about where to place blame. Sure, every driver has a duty to pay attention, and drive with caution - however, the sheer number of people injured by cars suggests there is a systemic problem. Blaming the driver in many incidents is like saying a lottery winner was skilled. The system is broken and has these injuries and deaths baked into it. Dealing with bad driving can only improve things to the limits of the systemic problems.


Reading "Let me try to explain using analogies.", I was worried someone would come up with another car anology.

Seeing orgs as people with characters, oh my, my eyes are opened now. Sounds weird but it makes sense the few times that I have worked on contracted projects.

For me, an added complexity in gauging the other side of the contract is that their personality changes depending on the actual people that are involved. When some higher-ups join the project it in later stages, the personality may turns even more mental than it already is.


Fascinating, like watching a Bob Ross video but he is drawing with SDF formulas as his palette. Very chilling and pleasant to watch.


BTW, does anyone know a good migration library for nodejs/typescript?

I was looking for one some months ago and kind of settled with Typeorm migrations but I don't like them that much. It feels clunky and doesn't seem to provide tooling for saving the schema. Little things that annoy me are the timestamp names for the migrations and the needless "extend this base class for your migration" Java cargo-culting which feels so weird in JS/TS.


Prisma is really great . I’ve used it for a year or so with resounding success.


Knex


We've used Knex and it works well provided you don't need features such as repeatable migrations.

Imo you should actually differentiate your migrations between schema changes and data mutations. That allows you to use a more powerful tool like liquibase/flyway for schema yet use your preferred language (JS/TS) for data mutations which might need other libraries.

I'm curious if people have tried this approach before.


Random: just don't use Bookshelf (ORM, based on Knex) unless you evaluate it well: when we used it, it did individual SELECTs for any included, related rows...


When using raw SQL in strings, I really miss the automatic formatting that is provided for HTML, TSX and TS with prettier.

Raw SQL query strings also do not compose well and I miss auto completion when writing them (yes, I'm a spoiled kid after so much Typescript usage).

As with everybody else, I didn't like existing ORM/builder approaches, so I built and use my own with type-inference: https://github.com/hoeck/typesafe-query-builder. Any feedback would be great because I have the gut feeling that this one has gone way too far on the type astronaut side of things.


You should be able to do this with IntelliJ e.g. where you can inject a language into a string. This is quite handy as you can reformat and open the string in a separate editor if needed.

The caveat is that it does not work well if you are composing the SQL queries from multiple small parts.

see https://www.jetbrains.com/help/idea/running-injected-sql-sta...

I suppose that other editors or IDE can do similar things.


Writing literal SQL in JS would IMO need more tools than just a preprocessor like you describe.

The few times I tried it (mostly in tests to check that the ORM is working properly) the #1 thing I was missing is a prettier-plugin that automatically formats SQL in the same way it currently works for `html` tagged templates.

I completely agree to the 'constrained by ORM' and 'useless features' part though. Postgres `json_agg` is a godsend and I love to be able to reason over simple joins and queries.

BTW, my own approach to use `json_agg`, `json_build_object` and json columns within a typesafe query-building DSL is this: https://github.com/hoeck/typesafe-query-builder

But its mostly for replacing simple ORM fetches, it wont do complex analytical queries. For that I'd like to write SQL directly as query-DSLs tend to quickly stop being usable in that situation.


One of my favorite features of WebStorm is the (official) database plugin which highlights SQL queries inside JS strings AND has autocomplete and refactoring support that actually uses the live database schema.


But unfortunately returned values have no type definition unless manually provided.


Tide + https://github.com/jscheid/prettier.el + projectile + helm work wonders. Been using this for +3 years now and its one of the bigger productivity improvements in my recent years of emacs and programming practice.

At work, I actually try to get ppl away from Jetbrains IDEs because they are slow and their typescript and prettier support seems pretty buggy to me.

Unfortunately, most developers I meet these times are not the kind of hacker types that want to understand and optimize their tools but rather just to get stuff done on the surface (which is totally fine). So emacs isn't just a good fit for them.


> prettier support

I use emacs/tide/prettier-js, but the prettier formatting only happens on save. Do you have settings that make emacs adhere to prettier formatting as it goes along?


Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: