Historians about history popularisers. Among ourselves we complain that often they give the wrong picture and provide no more information than reading Wikipedia, but we recognise they legitimise the profession and offer a gateway that is at least better than journalists and ideologues
Probably the oddest is King Henry VIII splitting from the Catholic Church. To people outside the Anglican Communion it was a much needed reform to break from a religious head who was letting temporal concerns affect spiritual decisions. Inside... Yeah, it's a dumb basis for a religion.
Excel. It's a genuinely amazing product and 90% of the complaints people have are skill issues-- but once you know enough, you should know enough to stop using it for most things.
Oh boy. The utility of rapid home tests for infectious disease. Accuracy is usually terrible (!), but people are a lot more likely to use them than something where they have to go submit a sample at a collection site.
There are useful applications in protein folding and meteorology (among others) but I see no complication in the chatbot LLMs at all. What's the complication there?
I don't know what you mean. Everything about it is complicated.
How they function and how well it can be explained. What they are or are not good at and in which contexts. How to connect them to external tools, and what's safe.
If they weren't climate-destroying plagiarism machines they would be somewhat good at entertaining and marginally useful as search engines in specific instances.
Almost all of what they are actually used for they are completely shit at.
They're quite good at writing code, which is hugely significant. That's far from the only thing they're good at, but that alone undermines this type of argument.
The tone of your post makes it clear that your position is primarily an emotional one.
They don't give answers: they give answer-shaped objects.
At that point you have to go and see if the actual answer matches the output. Only if the actual answer was particularly hard to find but easy to verify can they can have value.
Texas. Yes I want to gtfo because of the political climate and… literal climate, but I’m still from here and don’t want to listen to any “why don’t we just leave people in southern states to suffer, they deserve it” type bullshit from northerners.
1. They make it difficult for drivers to see bicyclists, which can be disastrous at conflict points (driveways that cut through the parking, intersections). For this reason they also don't help pedestrians as much, unless they also have daylighting. (More on this in a sec)
2. Parked cars only offer moderate protection as they're not attached to the ground. A speeding car can push a parked car into a bicyclist and still hurt them quite badly.
3. They're often seen as an "uncontroversial" solution because they preserve parking and pols rarely want to anger drivers by removing parking. But to do them right (e.g. with daylighting) you usually have to remove some parking. Hence, they invite cut corners (literally and figuratively).
4. In the worst case, if no cars are parked, they revert to unprotected lanes with a travel lane right next to them but worse, because now good actors will avoid that lane, and bad actors (e.g. people ok breaking laws) will use the parking lane as a high-speed passing lane, right next to bicyclists.
The condition of homeless shelters in my city. It's wholly insufficient and underfunded, but effort is being done and the people who work there are truly doing their best.
The company I work for. Talking trash publicly about your company is career suicide, and singing their praises looks better to future employers, helping you get the f- out of there!
As a research background psychotherapist, I think I just complain about them. Useful for population trends I guess, but dangerously misleading for individuals
Comments
To people who went full CBT => "sure it's not scientific, but let's not dismiss an useful way to explore the way we conceive stories about ourselves"
It's complicated.
How they function and how well it can be explained. What they are or are not good at and in which contexts. How to connect them to external tools, and what's safe.
They are much more effective when provided proper context (or the ability to retrieve it).
Almost all of what they are actually used for they are completely shit at.
The tone of your post makes it clear that your position is primarily an emotional one.
They are fast at producing code-like outputs that sometimes work and sometimes appear to work
At that point you have to go and see if the actual answer matches the output. Only if the actual answer was particularly hard to find but easy to verify can they can have value.
Internally, we all know what's going on.