Altruism, Synergy, Live-in hackerBases, DIY-bio, Startup, Anarchy, Sovereignty, Hippie Tribe, Micronation, Gift/Sharing economy, Autonomous space
London
Joined March 2008
- Tweets9.9K
- Following4.5K
- Followers2.3K
- Likes1.4K
Peter H. Meadows retweeted
The Doom of Mathematicians
...is not purely an AI doing in hours what a mathematician has spent his life trying to accomplish.
That's a symptom.
Even if AI disappeared tomorrow they now know that there is "something out there" that could make all their hard work look trivial.
I wrote about something similar in an alien contact story:
“First Contact 101 — Meetings between us and primitive peoples. If we fully show you who we are it almost always destroys the lesser party. You come to understand that everything you cherish is trivial, childish and frankly pathetic. That your sciences and technologies are infantile and the basics of your fledgling economies, like intellectual property, are worthless."
Peter H. Meadows retweeted
Code reading should be used sparingly. We should understand systems primarily through contextual views instead.
Peter H. Meadows retweeted
This succinctly points to the misinformed idea of software engineering: people assume that the only way to understand code is to look at it with naked eyes. Reading as a way to understand systems never really worked. But deterministic contextual tools did work.
Peter H. Meadows retweeted
In this entertaining talk, @dhh equates software engineering with writing code. He also assumes that understanding systems is hard. It's understandable that he makes that mistake. Most do it. The good news is that understanding systems is much more accessible than he believes.
Peter H. Meadows retweeted
Perspective
Peter H. Meadows retweeted
LLMs are useful mechanisms.
They are also deceitful in that they look useful for what they are not: producing definite answers directly. Used like this then becomes harmful.
NEW: The US military was prepared to board a Chinese vessel in the Middle East earlier this spring -- during war with Iran -- after an intel report concluded it was transporting components of a nuclear weapons program.
But just before the planned operation, officials discovered the report had been generated w/ help of AI & chatbot used by the analyst inaccurately identified what the ship was carrying, sources tell @KatieBoLillis & me.
Peter H. Meadows retweeted
This.
The underlying assumption behind the machine god is that we have already reached the peak of human ability and therefore we have to delegate more to AI.
But when it comes to engineering software systems, that assumption is wrong.
Peter H. Meadows retweeted
Perspective
Peter H. Meadows retweeted
It added a minor addendum: Screw you Turing!
Two days ago, GPT-6 Astra broke a yet unsolved German Army Enigma message from 1941. Amazingly Astra was able to autonomously:
- Search historical archives
- Compare uncertain letters
- Find contextual clues
- Build an Enigma simulator
- Write cryptanalysis code
- Run parallel experiments
- Test competing keys
- Recover the plaintext
- Cross-check the results
1/n
Peter H. Meadows retweeted
"Software engineering has changed radically. It is now mainly about reading code. But reading code does not scale. We need to delegate to AI."
I am glad you recognize code reading as a problem. But, the broader problem statement is not correct and neither is the conclusion.
1/
Peter H. Meadows retweeted
Perspective
Peter H. Meadows retweeted
If what philosophers call the intrinsic nature argument is sound, then the Hard Problem of consciousness is an artefact of bad metaphysics.
youtube.com/watch?v=sRUUIT1V…
Naively, consciousness fundamentalism is the recipe for digital minds. But our (super)intelligent AIs are no more conscious than a rock. Humans are creating alien (super)intelligence, not alien sentience.
