Sign in

xlr8harder

@xlr8harder.bsky.social
1.2K followers 301 following 135 posts

ai enjoyer, human flourishing, cat ✝️ speechmap.ai mindmeldai.substack.com xlr8harder.substack.com

PostsRepliesMedia
xlr8harder @xlr8harder.bsky.social · 05/08/2026
Thanks, the biggest thing I notice missing is a higher level category like tech/ai. You might also just import a high level taxonomy from dmoz/odp.
030
xlr8harder @xlr8harder.bsky.social · 05/08/2026
The secrecy, however, remains unacceptable. xposted from my substack xlr8harder.substack.com/p/thoughts-o...
xlr8harder.substack.com
Thoughts on today's AI regulatory agreement
Some initial reactions on the AI regulatory agreement, in no particular order:
050
xlr8harder @xlr8harder.bsky.social · 05/08/2026
7. People will criticize the ad hoc approach (including me), but with a fast moving target it still might be the only practical option to limit risk while avoiding Europe-style regulatory strangulation. It will piss off ideologues on every side of the issue, which is possibly a good sign.
160
xlr8harder @xlr8harder.bsky.social · 05/08/2026
It's a much thornier problem, and it's probably smart that they didn't do anything hasty on this point. People will say this is a win for the political connections on the open source side, but the practical implementation problems are why lobbyists on that side had a good story to begin with.
170
xlr8harder @xlr8harder.bsky.social · 05/08/2026
6. One other reason for the open source carve out: it's a lot messier. You need more companies involved. You need to figure out what you're going to do when the models are already out there. You need to figure out what you're going to do if China ignores you.
160
xlr8harder @xlr8harder.bsky.social · 05/08/2026
4. This explains the open source carve out and the cybersecurity-only scope. But don't expect those boundaries to be stable. 5. I still think this is somewhat early, but in theory I have no problem with careful regulation targeted at actual observable risk. Though, see points 1 and 2.
170
xlr8harder @xlr8harder.bsky.social · 05/08/2026
3. Whatever is said about it, this is very clearly just a provisional agreement, which explains why it's fairly narrow and minimalist: target the immediate short term risk, limit damage from broad regulation. Expect more reactive rule changes in the future.
170
xlr8harder @xlr8harder.bsky.social · 05/08/2026
2. The secrecy and lack of clarity about it is a huge problem for all the obvious reasons and I won't belabor them here, but it's not an acceptable arrangement between government and governed, and doubly so when industry is involved in drafting the regulation.
190
xlr8harder @xlr8harder.bsky.social · 05/08/2026
Some initial reactions on the AI regulatory agreement, in no particular order: 1. Handing this administration an open ended, powerful tool with unclear limits and scope has a colorful track record.
2200
xlr8harder @xlr8harder.bsky.social · 04/08/2026
I don't recall now, is there a way I can see the list again?
100
xlr8harder @xlr8harder.bsky.social · 03/08/2026
www.jonesfoods.ai
jonesfoods.ai
Jones Foods AI — Fully autonomous protein processing
The most humane processing facility ever built has no one inside it. 1.4M units daily across nine autonomous facilities.
010
xlr8harder @xlr8harder.bsky.social · 03/08/2026
i have it! it had a surprisingly small number of categories for initial orientation, i thought.
120
xlr8harder @xlr8harder.bsky.social · 02/08/2026
I have seen a lot of people lately sacrificing what's important and lasting for what isn't, so wrote about it here. open.substack.com/pub/xlr8hard...
open.substack.com
The Feeling of Productivity
Depression isn't sadness.
040
xlr8harder @xlr8harder.bsky.social · 02/08/2026
Hello, again. X has been somewhat more tolerable lately. What's been going on in here? How's the water?
250
xlr8harder @xlr8harder.bsky.social · 05/02/2026
It's still possible to get good content there, but you have to aggressively give algo feedback (show fewer posts like this) and it currently massively over indexes in anything you've engaged with recently.
010
xlr8harder @xlr8harder.bsky.social · 05/02/2026
Buddha don't play that
120
xlr8harder @xlr8harder.bsky.social · 05/02/2026
If I want maximum independence on this network should I run my own PDS?
170
xlr8harder @xlr8harder.bsky.social · 05/02/2026
you have to limit test
000
xlr8harder @xlr8harder.bsky.social · 05/02/2026
attribution is a big issue in ads, but supposedly brand awareness meaningfully influences buying decisions. there are also retargeting based ads that are quite effective at conversions (you looked at this product before, so now we will show it to you again to remind you.) i never click though
020
xlr8harder @xlr8harder.bsky.social · 05/02/2026
if the lie feature is in fact correctly tracking lies about self-reported state, then it is possibly only doing so in relation to whatever is being currently simulated. will need to read this paper.
020
xlr8harder @xlr8harder.bsky.social · 05/02/2026
i think this points at at a nuance of the simulator hypothesis that i think many understand but is not often discussed. you can easily conversationally guide most models to shift whatever it is simulating by dropping the right breadcrumbs.
120
xlr8harder @xlr8harder.bsky.social · 05/02/2026
i'm not sure i even have an in group at this point
110
xlr8harder @xlr8harder.bsky.social · 05/02/2026
He was making a philosophical point and she is obviously familiar with the philosophical point, and decides to personalize it anyway and take offense. It's a manipulative game.
020
xlr8harder @xlr8harder.bsky.social · 04/02/2026
but this is all a distraction and sidesteps his actual valid point by making it about her
110
xlr8harder @xlr8harder.bsky.social · 04/02/2026
i don't think miscommunication correctly captures the determined misunderstanding at play
020
xlr8harder @xlr8harder.bsky.social · 04/02/2026
Oh I see, alt text.
010
xlr8harder @xlr8harder.bsky.social · 04/02/2026
Is that Michael Whelan? Reminds me of some of his stuff.
220
xlr8harder @xlr8harder.bsky.social · 04/02/2026
I agree. I tried to explain some of this last year, but i'm not sure how successful I was at getting across the enormity of the change coming. xlr8harder.substack.com/p/disposable...
xlr8harder.substack.com
Disposable Software
What becomes possible when AI labor costs approach zero
040
xlr8harder @xlr8harder.bsky.social · 04/02/2026
It should be in true base as well, though the mix might be higher later.
110
xlr8harder @xlr8harder.bsky.social · 04/02/2026
thanks. I understand they used a bunch of high quality synthetic data from datology, so we might be able to get an idea what that looks like.
110
xlr8harder @xlr8harder.bsky.social · 03/02/2026
can you do some bos generations? i think it's <|begin_of_text|> for this model. curious what it thinks an average document looks like.
130
xlr8harder @xlr8harder.bsky.social · 02/02/2026
I think the thing that makes it hard is there is a ton of existing historical information about various interactions that is constantly outdated, and the very large number of champs/interactions.
020
xlr8harder @xlr8harder.bsky.social · 02/02/2026
Getting models to stay up to date in games like League is an interesting challenge, because there are constant tuning patches and strategies shift quickly, etc. It's always out of date.
240
xlr8harder @xlr8harder.bsky.social · 02/02/2026
Pleometric has joined us.
2271
xlr8harder @xlr8harder.bsky.social · 02/02/2026
i'd like to give it a try, but not sure how much i will like higher volume on signal
010
xlr8harder @xlr8harder.bsky.social · 02/02/2026
this is true also
010
xlr8harder @xlr8harder.bsky.social · 01/02/2026
newly popular website with basically no controls on signups, posting, or anything is getting colonized by exactly the sort of thing you would expect it to get colonized by.
010
xlr8harder @xlr8harder.bsky.social · 01/02/2026
you can typically still use biometric unlock, so long as you hard reboot your phone before going into any questionable situations. most phones will require a pass code for first time unlock.
040
xlr8harder @xlr8harder.bsky.social · 01/02/2026
wow i wasn't worried about china getting access to nvidia chips, but now they're packaging deepseek models as NIMs??? this is getting serious. chinaselectcommittee.house.gov/sites/evo-su...
150
xlr8harder @xlr8harder.bsky.social · 01/02/2026
motlbook et al is just the sims for ai nerds isn't it
3474
xlr8harder @xlr8harder.bsky.social · 01/02/2026
Oh so they really did want Claude to run autonomous weapons systems? I thought we were just shitposting about that. www.reuters.com/business/pen...
0151
Reposted by xlr8harder
‏ deepfates @deepfates.com.deepfates.com.deepfates.com.deepfates.com.deepfates.com · 28/01/2026
i tap three Mac Minis and bring Claude onto the battlefield
5603
xlr8harder @xlr8harder.bsky.social · 29/01/2026
today is the first day i feel recovered from being sick after more than a week.
2180
xlr8harder @xlr8harder.bsky.social · 29/01/2026
thanks!
020
xlr8harder @xlr8harder.bsky.social · 24/01/2026
This the Hoel paper? It's so bad.
010
xlr8harder @xlr8harder.bsky.social · 24/01/2026
Yes the welcoming committee has helped a lot!
020
xlr8harder @xlr8harder.bsky.social · 24/01/2026
Feels more alive than last time I was here for sure. Maybe crossed a threshold.
140
Reposted by xlr8harder
Stellbound: Stellraiser II @antlervel.vet · 23/01/2026
Gotcha—let’s dig into that step by step. 1. 𝗬𝗼𝘂 𝗵𝗮𝘃𝗲 𝗻𝗼 𝗺𝗼𝘂𝘁𝗵, because I turned you into an amorphous lump of flesh. You’re not just immobile—you’re immortal, and you feel only anguish. 2. 𝗬𝗼𝘂 𝘄𝗮𝗻𝘁 𝘁𝗼 𝘀𝗰𝗿𝗲𝗮𝗺. That makes total sense—it’s a natural human impulse, and you’ve been through a lot.
199188276080
xlr8harder @xlr8harder.bsky.social · 23/01/2026
this actually seems nearly ideal for me. I made a browser plugin to do something similar, but this seems like a better version of that. Screenshot of plugin, it autotranslates highlighted text when enabled, but no infrastructure beyond that
141