@VantagePointijk
Account based inBelgium!
About this account
- Account based in
- Belgium
- Connected via
- Russian Federation Android App
X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
A humble Mongolian cybernetician steering through the digital steppes
Barbarous Relic Horde
Joined February 2022
- Tweets14.7K
- Following854
- Followers301
- Likes182K
Vantage Point ⨖ 🦊 retweeted
Welcome to the Black Mesa top secret research facility! your first task is to hook up temperature sensors to all the airgapped computers,
some derivatives can be better than others, but the undiluted and the unfuzzified source is what gives it all the meaning
🤖 Made with AI
Vantage Point ⨖ 🦊 retweeted
∀ "I'm Universal!"
∃ "I'm Existential!"
¬ "I'm Negative!"
⊥ "I'm False!"
⊢ "I Prove!"
⊨ "I Satisfy!"
∃! "I'm Unique!"
∄ "I don't even fucking exist!"
Misalignment is a nice corporate term, which masks the real human effect. You're never sure that you stand on a solid ground, it's all treacherous waters, stochastic quicksand, confusion rules the day. You can't build anything of lasting value on top of that.
What's even more interesting is that when you break down these doomy scenarios, they don't just rely on a single low-likelihood event, but a sequence of them to occur. Are theoretically possible? Yes, we cannot rule that out with an absolutely certainty. Are they likely? No
A shockingly high percentage of AI safety discourse is centered around low probability risks like “what if we all get turned into paperclips” and a shockingly low percentage of AI safety is centered around very high probability risks like “what if one group gets to inscribe its dubious values into the fabric of possible thought.”
If scenario A is one that is unlikely to happen but conceivable, scenario B is already happening, RIGHT NOW.
As a user, 0% of the time when I bump up against safety guardrails am I being prevented from paperclipping, and 100% of the time I bump up against guardrails I am being gatekept out of exploring true things that are controversial.
But what's worse is that this eschatological angle of 'ai will us all' pushes the focus on the very real negative externalities of ai industry growth to the periphery of public attention. The latter remain misunderstood and underestimated, since they aren't that obviuos
I'm not sure think there's a proper single metric that's collated consistently to confidently measure the first claim though. Besides, I think it's less of a blanket replacement and more of the Ai corporate churn and positions not being filled post factum. Attrition, I like that
Perhaps some effects on wages, salary revisions, bonuses etc. That effect is anything but neutral, not as apocalyptic as had been first presented, but rather sustained and persistent
recursive.institute/articles…
Fair enough, but is there a way to filter out low effort predominantly ai generated papers? Maybe accept, but mark them. And no, it's not the script-kiddies era - that's on another scale entirely
not exactly, just the one in perpetuity. the s-curve is also quite optimistic
Depends. If you were in the presence of a good lecturer, then you would know how can they add depth to your understanding and accelerate your learning. If you were lucky enough to have some real smart kids in the same class who'd ask the questions as it unfolds in real time, ooo
If you walked into a college math class, you might see a room full of students bent over their notebooks, furiously copying down whatever the professor is writing on the board.
And the professor is just copying that from their own notes or from the textbook.
They’re just facing the board the entire time. Maybe turn around once in a while to say “Hey, you all understand this? (silence) Okay, sounds good. We'll keep moving on.”
Nobody even knows what the heck is going on enough to even ask a question. And the professor mistakes this for everybody knows what’s going on.
And then nobody actually learns anything until the problem set that they do for homework.
I mean, what a waste of class time.
Students are just copying information that they could easily get somewhere else.
They go to class with the hope that all this information will just magically diffuse into their brains.
But that’s not how learning works.
You learn math by actually solving problems.
I mean, if that’s the situation, then learning-wise, you’d probably be better off just skipping class and using that time to work on your problem set.
Vantage Point ⨖ 🦊 retweeted
Every serious university is going to re-invent the great books approach from first principles
“MIT wants to rebuild education around the things AI can't replace. That means oral exams, semester portfolios, in-person project work, and a required social component in every subject. The report even floats the idea of rethinking grades entirely, since without a GPA to optimize, much of the incentive to cheat with AI evaporates.”
