I've done something like this when writing grant applications, etc. When space is at a premium, strategically shortening/deleting words to eliminate a widow can gain you a whole line :)
> GPT2 was considered [possibly] too dangerous to release
Fixed that.
> Everything is "too dangerous". GPT2 is going to invent a time machine and break crypto and genetically engineer super rabies.
You are writing this with knowledge and experience they didn't have.
Judging a spectrum of responses, based on a subset of that spectrum, after the fact, doesn't have much utility.
On the other hand, there is almost certainly greater utility in outsized caution than not, when playing with cognitive technology that is rapidly gaining ground. Where "rapid" here means "many orders of magnitude faster than human evolution".
Whereas I made a mistake and bought an underspeced MacBook and my 512GB disk is constantly on the brink of full as I reclaim the last 40 GB over and over from different caches, downloads, and wherever else and had to simply give up on several projects because of their disk usage.
Obviously anything unhappy that happens anywhere in the world is because of Israel. If you stub your toe, it’s probably on something a Jew placed there somehow.
That, or there’s been a naturally shift to the right as a reaction to the left going to irrational extremes and aligning themselves hypocritically with groups that suppress women and hate the USA and the freedoms we have fought for and won.
What's missing from these discussions is the real weakness of diffusion models: you can have two positions where two tokens need to be coordinated, but both spots are teetering between two valid states. They don't always collapse to the matching pair.
I only learned this the hard way reimplementing diffusiongemma. I had ideas on how to fix it but no cluster to train and experiment, hah.
I love this. Carney called a fight, Trump just throws a tantrum.
That guy began putting his name in tall buildings, in big golden letters. Then he wanted to have his name in the Kennedy center. Then he "changed" the name of the Gulf of Mexico. Now this.
It is all typical Trump: narcissistic, theatrical, childish, and most of all, irrelevant.
If he were to be relevant and consequential he would do what he can't: raise even more taxes on other Canadian imports (beyond the 5% he taxed now) or even threaten the country militarily. But he can't do these things when Americans voters complain about inflation and war 2 months before an election.
He is weak and scared.
This will be a long fight, but so far Canada is winning.
Much of your argument rests on the claim that those terms are fully interchangeable. Many posters, both native and non-native speakers, have observed that they are not fully interchangeable.
I don't remember how it happened, but I have seen faulty Ethernet devices suffering a power surge that burned the switch port they were connected to, so be careful.
> Choosing words that carefully takes some serious dedication.
No it doesn't. If you're careful about how you speak and how you write then
it takes next to no additional effort at all. I used to do this with all of
my online responses, and it's been fun to do it again in this thread today.
Note my responses in this comment thread as a whole. Even though I have not
done this more than once or twice this century, it barely slows me down. At
a guess, maybe it adds 10% or so to the time it takes to type a comment...?
> Imagine the backtracking required...
Virtually none. For some lines I might occasionally have to go back a three
or four words, that's all. It's quite fun: some dusty rusty old bits of the
old grey matter have been woken up and exercised, and they're rather liking
being up in the sunlight again. :-)
They are different like you say, but the jump from an interactive, agentic shell like pi to a long running agent like OpenClaw is not a hard jump to make with extensions and scripts. Even just running pi in a cron job will get you a large portion of the way there.
I get that the guy needed to write a paper on something to get his PhD (in psychology, which might explain some of this as well, I thought standards were a little higher in the hard sciences), but don't know why it deserves a high profile. At least, based on merit.
