It's the age of LLMs! Language has been solved! LLMs are great at both Czech and Polish. This problem is orders of magnitude easier. Why doesn't my keyboard even know these words exist?? Is there an Android keyboard that actually... knows basic forms of basic words?
Really? If you swipe "kill" and then try "yourself" or "myself" does it ever get it right or provide it as one of the options? Doing it right now myself and I can't get it to do either. I have manually entered those words and hit the "myself" in the suggestion box to try and convince it that that's an acceptable correction to no avail.
> I inevitably do the "wrong thing" and fall victim to the editing again, or tap something wrong, or.. I don't know
Every. Time. I like to think that I'm not an idiot and can generally pattern recognize, but it just feels so inconsistent that I'm always doing the wrong thing.
Is this some sort of psyop to get me to use siri to send texts?
This is absurd to the point of being cartoonish. No one treats billionaires like supervillains. How many billionaires are in supermax prisons right now in New York?
> Nobody particularly likes them
This is not relevant, regardless of whether it’s true. A ton of people hate Thiel and Trump. Disliking a billionaire doesn’t take away their power.
"Supervillains" are comic book entities who are rarely in prison
1) LLMs have failed to live up to the hype.
Maybe. Depends upon's who's hype. But I think it is fine to say that we don't have AGI today (however that is defined) and that some people hyped that up.
2) LLMs haven't failed outright
I think that this is a vast understatement.
LLMs have been a wild success. At big tech over 40% of checked in code is LLM generated. At smaller companies the proportion is larger. ChatGPT has over 800 million weekly active users.
Students throughout the world, and especially in the developed world are using "AI" at 85-90% (from some surveys).
Between 40% of professionals and 90% (depending upon survey and profession) are using "AI".
This is 3 years after the launch of ChatGPT (and the capabilities of chatGPT 3.5 were so limited compared to today that it is a shame that they get bundled together in our discussions). I would say instead of "failed outright" that they are the most successful consumer product of all time (so far).
I have a really hard time believing that stat without any context, is there a source for this?
In the T-shirt example if you left the program as-is but then decided that tackling abuse is a separate topic, think about what that would look like: Every maintainer would now not only have to read and close the spam PRs, they’d have to go file an abuse report for every single one of them. Now you’ve put even more work on the maintainers and created an additional burden of reviewing reports, all without clarifying the program to discourage abuse from the start.
This is why it’s necessary to structure a program clearly such that abuse-level or low effort inputs can’t easily claim the rewards.