David Corfield retweeted
I'm doing a course based on my Cultures of Programming book! A perfect excuse to also record the material for anyone who prefers listening over reading.
Part 1: What can we learn from the 80 years of programming history & why Knight Capital lost $440m:
youtube.com/watch?v=Kjf6Mi…
Part 1: What can we learn from the 80 years of programming history & why Knight Capital lost $440m:
youtube.com/watch?v=Kjf6Mi…
4
14
81
3,680
Funny how enquiries that appear to belong to rather different spheres end up intersecting. Currently for me this is control theory meets psychodynamics. So in my day job I'm part of a team investigating by category-theoretic means ways to formulate safety conditions and...
1
0
4
419
There's perhaps a @Mark_Solms take on this, which would have barrier functions fuelled by FEAR, and the need to have obstacle-removing RAGE do its work. However before this can happen, the further FEAR-driven barrier against using RAGE to do this, which has led the latter...
1
0
0
129
to turn in on the psyche, needs dismantling. RAGE needs to find its proper object in the carer who let one down.
The categorical control theory is fascinating too.
The categorical control theory is fascinating too.
0
0
0
113
David Corfield retweeted
Funding has just been renewed, for the next 5 years:
ncatlab.org/nlab/show/Cent…
ncatlab.org/nlab/show/Cent…
4
6
76
2,701
David Corfield retweeted
I’ve recently accepted the Programme Director role at ARIA, taking over from davidad in running the Safeguarded AI programme.
🧵 about the programme’s strategic vision + our upcoming funding efforts in cybersecurity + we're hiring!
🧵 about the programme’s strategic vision + our upcoming funding efforts in cybersecurity + we're hiring!
11
24
233
19,086
David Corfield retweeted
A new web page for my book! tomasp.net/cultures/
It has all the links you need to get the open access PDF, buy a hardcopy or an ebook, as well as some older talks.
If you run a podcast, conference or a user group, I'm always hapy to join & talk about something from the book!
It has all the links you need to get the open access PDF, buy a hardcopy or an ebook, as well as some older talks.
If you run a podcast, conference or a user group, I'm always hapy to join & talk about something from the book!
2
13
93
4,200
Looking forward to a happier post-postscript.
The open access version of Cultures of Programming is now available online. Just in time for the holiday break!
Read it here: cambridge.org/core/books/cul…
If you prefer a real book, use PETRICK26 for a 20% discount (valid until January 31) at: cambridge.org/9781009492348
0
0
4
559
There's something of Gendlin et al.'s 'Think at the Edge' to this: focusing.org/gendlin/docs/g…
A paper that seeks to extract concepts from an AI chess player that are currently beyond human experts, but are still teachable:
pnas.org/doi/10.1073/pn…
Having extracted the most powerful novel AI concepts as vectors, they look to see if these are outside the human range
0
1
4
970
Converging ideas, arxiv.org/abs/2505.03335:
Some reflections here (ncatlab.org/davidcorfield/…) on the extent to which may see Charles Peirce's three modes of reasoning (deduction/induction/abduction) as completions of inferential triangles.
0
1
14
977
"opposite poles of reason", i.e., dual constructions: ncatlab.org/davidcorfield/…
Abduction and induction have, to be sure, this common feature, that both lead to the acceptance of a hypothesis because observed facts are such as would necessarily or probably result as consequences of that hypothesis. But for all that, they are the opposite poles of reason,
/3
1
0
8
705
Some reflections here (ncatlab.org/davidcorfield/…) on the extent to which may see Charles Peirce's three modes of reasoning (deduction/induction/abduction) as completions of inferential triangles.
0
1
19
2,105
A paper that seeks to extract concepts from an AI chess player that are currently beyond human experts, but are still teachable:
pnas.org/doi/10.1073/pn…
Having extracted the most powerful novel AI concepts as vectors, they look to see if these are outside the human range
pnas.org/doi/10.1073/pn…
Having extracted the most powerful novel AI concepts as vectors, they look to see if these are outside the human range
1
8
21
3,055
"We define the novelty score based on how well a concept vector can be reconstructed using a set of basis vectors derived from AZ’s games. A lower reconstruction loss means that the concept is better represented by the given set of basis vectors...
2
0
4
419
...In other words, we look for concepts that are better explained using AZ’s language (basis vectors) than humans’ language."
So better bases allow the easier expression of powerful concepts. Interesting.
So better bases allow the easier expression of powerful concepts. Interesting.
0
0
3
379
I'd love to see this vector space approach meet with dependent type theory. There should be a way to pass back and forth between the hierarchical encoding of dependent typing and corresponding representations within a very high-dimensional space.
Fundamentally, high-level concepts group into categorical variables---mammal, reptile, fish, bird---with a semantic hierarchy---poodle is a dog is a mammal is an animal.
How do LLMs internally represent this structure?
arxiv.org/abs/2406.01506
3
1
21
1,465
Interesting. They give a h/t to @_julesh_ who had related laxness of functors to emergent effects (julesh.com/2019/12/02/lax…).
Maybe we can get category theory to speak to other approaches to emergence, such as @erikphoel et al. arxiv.org/abs/2104.13368, arxiv.org/abs/2202.01854.
Maybe we can get category theory to speak to other approaches to emergence, such as @erikphoel et al. arxiv.org/abs/2104.13368, arxiv.org/abs/2202.01854.
Caterina Puca, Amar Hadzihasanovic, Fabrizio Genovese, Bob Coecke: Obstructions to Compositionality arxiv.org/abs/2307.14461 arxiv.org/pdf/2307.14461
1
2
19
2,480
it's mono.
If so, what do departures from isomorphism look like in this category? Do they relate to what Comolatti and Hoel have on determinism and degeneracy as generalizations of sufficiency and necessity?
If so, what do departures from isomorphism look like in this category? Do they relate to what Comolatti and Hoel have on determinism and degeneracy as generalizations of sufficiency and necessity?
1
0
3
534
I see @erikphoel has now issued a Causal Emergence 2.0, arxiv.org/abs/2503.13395. May be worth revisiting this thread.
0
0
0
267
Quiver representations look very interesting here: arxiv.org/abs/2207.12773
But taking the first steps to go symmetric situations, are there people looking to utilise what's coming from monoid representation theory? This is notorious less simple (mathoverflow.net/q/37115/447), but there is material on the subject, such as ...
0
0
5
660