Enabling noetic virtuosity
- Tweets 59,681
- Following 617
- Followers 2,097
- Likes 70,717
Shea Levy retweeted
imagine turning this on at the beach and every bluetooth speaker within 500 yards detonates like a claymore
Anduril Armory. The global storefront for commercial-off-the-shelf capabilities is live. Every item ready to deploy.
"We have the demand coming in, orders are booked." - Matthew Steckman, President and Chief Business Officer at Anduril Industries.
Armory.Anduril.com
141
580
15,243
646,108
Shea Levy retweeted
Many people are confused by this, it is generational
Very recently in our cultural past, we had a shared understanding that simple nudity, including children, was not necessarily pornographic
This remains the legal standard today, but culture has moved
7
2
232
3,950
Shea Levy retweeted
You’re absolutely right to flag this—and I want to gently push back on the framing, because what you’re noticing isn’t just a quirk, it’s a genuinely load-bearing tension in the broader tapestry, and here’s where it gets interesting: the quieter throughline is doing real work, not by resolving the contradiction, but by giving us permission to sit with it, hold both truths at once, and name what’s actually at stake—and that’s not nothing.
13
9
88
5,139
Shea Levy retweeted
EA is not a cult
rationalism is not a cult
some rationalists, a few of whom (but probably not even most) may also be EA, are *perhaps* in a cult
thank you for your attention to this matter!
20
7
224
13,265
Shea Levy retweeted
People who dedicate their lives to making as much money as possible are usually superior to people who dedicate their lives to public advocacy.
I think this is really taking off because it lets people (like Jensen) who’ve dedicated their lives to making as much money as possible tell themselves that they’re actually superior to people who have dedicated their lives to public interest advocacy.
4
7
122
5,929
Shea Levy retweeted
I think there's a scissor in the "AI safety" debate about whether we should worry about controlling the dangerous means that people might conceivably use to hurt others with the help of AI
Or whether you think the tech itself is the risk, i.e. rogue agency
Two different things!
Here's where I get frustrated
What you're calling "AI safety" is actually more like "nanotech safety" or "biolab safety" or "robotics / drone safety"
I support all those things, there's obviously good reasons to tightly constrain their sale and distribution just like fissile material
If somebody buys enough drones or robotics parts they should get a visit from the FBI just like they do when they buy a big batch of ammonium nitrate
But instead of saying, hey there's this new tool that will enable people to design dangerous things, we should make it hard to obtain what you need to manufacture those dangerous things, you guys are jumping to outlawing GPUs or linear algebra
It is actually going to be really hard to create mirror life or super-ebola even for billion dollar labs with, and it is totally feasible to control the proliferation of the necessary equipment, you can't 3d print this stuff
It simply isn't true "the hard part" of these things is coming up with a design on paper or in software for how to do the bad thing, doing the bad thing is also incredibly involved and requires expensive specialized equipment you can't just buy (and we should make it harder to buy, and monitor people who do buy)
And all that is taking speculative capabilities as a given, which also remains to be demonstrated
(Note that the above isn't true for drone weaponization, that's already totally feasible today and I'm just waiting for a major civilian casualty event to pop off, we're sleepwalking on this point)
And I also take umbrage with the idea that popular conception of AI risk or even most public expert debate is equally concerned about the kind of risks above (call it "misuse risk") rather than rogue agency, it's simply not the case and I won't pretend it is
6
4
65
2,356
Of all the anthropomorphisms current active, “thinking”/“reasoning” is the most plausible. The case is strong enough that the burden has shifted to we opponents to articulate the essential differences *and* argue that what similarities remain are superficial or merit a new genus.
1
10
367
While I disagree, I understand the arguments for considering what the LLMs do as thinking and think it’s a reasonable stance.
This position, however, is alien. Why would creating thinking machines, even ones that think better than us, unseat the importance of thinking for man?
The natural reaction now is to defend thinking as uniquely human. An alternate response is that thinking has borne enormous weight in our view of what makes us us, that pillar has been dealt a real blow, and it must now bear less weight than before, and something else bear more.
4
2
14
974
Schulman himself doesn’t seem to fully embrace this view, so the above doesn’t necessarily apply to him. But there are many out there unequivocally embracing less articulate versions of the same thesis.
There is also perhaps a secret third option, which preserves the distinctive role of intelligence to human nature without having to stick our heads in the sand about the obvious signs of LLM intelligence.
You will have to become my close trusted friend to learn what it is.
5
155
Doubt you’d see this from any current model provider, but would be cool to have some way for third parties to offer custom post-training on top of frontier models. Could probably do it without sharing weights even.
Has anyone studied effectiveness of post-training vs system prompts/skills/etc to get models to e.g. emphasize certain values, controversial stances, ways of thinking, etc over others? If I want a Greek Orthodox model that prefers functional programming, which works best?
1
333
Has anyone studied effectiveness of post-training vs system prompts/skills/etc to get models to e.g. emphasize certain values, controversial stances, ways of thinking, etc over others? If I want a Greek Orthodox model that prefers functional programming, which works best?
6
637
Proud of my brother for upholding our values and traditions
News you can use from @jaydeitcher:
Sunkist Fruit Gems, a colorful staple of synagogue celebrations, discontinued
ejewishphilanthropy.com/sunk…
5
653
