John Grant pfp

John Grant

@jlg

118 Following
166 Followers
John Grant pfp
John Grant
@jlg
How long before LLMs, particularly those trained on synthetic data, start to label the field of anthro-complexity as 'Nuancology'?
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Problematise: Abstract optimism without a roadmap is just wishful thinking.
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Interesting read. But the r2 values 0.32~0.44 indicate there's something amiss, right? Might suggest that human-AI interactions aren't following deterministic patterns. They're emergent and context-specific instead https://www.microsoft.com/en-us/research/publication/the-impact-of-generative-ai-on-critical-thinking-self-reported-reductions-in-cognitive-effort-and-confidence-effects-from-a-survey-of-knowledge-workers/
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
"… the persistent weirdness of our times. One in which we get free AI from a hedge fund and $200 a month AI from a nonprofit …" https://www.youtube.com/live/zdy9K4UWplM?si=HUN6Qb6Fx3UWFzBG&t=1164
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Anyone familiar with the Strategy Evaluation Protocol (SEP)? I'm looking for some pointers or examples of how SEP is currently being used to support Open Science.
0 reply
0 recast
1 reaction
John Grant pfp
John Grant
@jlg
From a certain perspective, Open WebUI looks like a mashup between a terminal emulator and a web browser. But I suspect there’s more to it. I think this might be a nascent form of what the next generation of operating systems will look and feel like.
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
“OpenAI's o3 finds a polynomial time factoring algorithm, none of your crypto assets are safe” https://x.com/realgeorgehotz/status/1881142120171626867
1 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
I'm finding Jensen Huang's presentations more entertaining than recent Netflix sci-fi output. https://www.youtube.com/watch?v=k82RwXqZHY8
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Should Elon Musk succeed in triggering a sovereign debt crisis, he's well positioned to lobby for X to be the default CBDC platform.
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Any sufficiently reductionist explanation is indistinguishable from superstition.
0 reply
0 recast
1 reaction
John Grant pfp
John Grant
@jlg
>"Our collective performance last year was not among our best, with five wrong answers. Following polling at the time, we discounted Trump’s re-election chances. Japanese interest rates did rise above zero; investors didn’t move back into bonds as expected; X didn’t go bankrupt; and there was no deal, yet, to return the Parthenon marbles to Greece. Single minds, it seems, can sometimes beat 20: two entrants to our reader competition got all 20 questions right, and the overall winner, Ercole Durini of London, was even spot on with the tiebreaker." https://www.ft.com/content/ac44e3a5-36ee-4cf8-af57-06a1ba51baa4
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
>"But what would be necessary to have your “own Cloudflare” thing you can run once and then run all your cool mini CGI like scripts above? We miss some necessary protocols. Yet building these protocols is tricky because you target often the least common denominator. ... Don't need any new innovative technology here, but you need the social contract and the mindset. Those are hard things, they require dedication and marketing. I have not yet seen that, but I'm somewhat confident that we might see it." https://lucumr.pocoo.org/2024/6/26/what-is-self-hosted/
0 reply
0 recast
3 reactions
John Grant pfp
John Grant
@jlg
>"… the economy itself emerges from our subjective beliefs. These subjective beliefs, taken in aggregate, structure the micro economy. They give rise to the character of financial markets. They direct flows of capital and govern strategic behavior and negotiations. They are the DNA of the economy. These subjective beliefs are a-priori or deductively indeterminate in advance. They co-evolve, arise, decay, change, mutually reinforce, and mutually negate. Subject and object cannot be neatly separated. And so the economy shows behavior that we can best describe as organic, rather than mechanistic. It is not a well-ordered, gigantic machine. It is organic. At all levels it contains pockets of indeterminacy. It emerges from subjectivity and falls back into subjectivity." This was written by Brian Arthur in 1999. Over the past 25 years what evidence either reinforces or challenges this perspective?
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Reading Guy Standing's The Politics of Time. Just as LLMs can simultaneously produce "coherent results" and act as "bullshit generators" (using Booch's framing), the UK's furlough scheme amplified both protection and inequality through its interaction with technology. Digital technology enabled knowledge workers who could work remotely to benefit disproportionately, while essential workers continued at personal risk. Asymmetric amplification might help explain how tools like LLMs or policies like furlough schemes interact with existing social dynamics in that they don't just serve dual purposes, they systematically amplify advantages for some while constraining or disadvantaging others >"The UK's furlough scheme… was possibly the most regressive social policy in modern history. It paid 80 per cent of the wage of employees up to £2,500 a month. This meant that somebody earning £3,000 beforehand received £2,400 for doing nothing, whereas somebody who had been earning £800 received £640." 1/n
1 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
"… LLMs are pretty freaking cool. However, they are, by the very nature of their architecture, unreliable narrators, that's what I say politely. If I'm going to be impolite, I will say they allow us to build at global scale bullshit generators, because they are clearly stochastic parrots. They do not reason they do not understand but they do produce some very coherent results because they allow us to navigate a latent space that has been made very complex through training it through the corpus of the internet." - Grady Booch This has always seemed rather obvious to me, but I don't find it easy to articulate. An axe can be both a survival tool and a lethal weapon. So, it can be said to be dual-purpose. But I find this analogy not as effective when applied to generative AI. In human systems, what framework or concept best explains this nature of things that can produce both beneficial and harmful effects? Are they examples of emergent properties, or is there a better way to frame it?
1 reply
0 recast
2 reactions
John Grant pfp
John Grant
@jlg
Pareidolia describes pattern-seeking behaviour in individuals. Is there a term for when organizations demonstrate collective pattern-seeking behaviours similar to pareidolia? I'm not sure it's the same as institutional confirmation bias – that would be selectively interpreting information to support existing beliefs. Pareidolia is spontaneous pattern recognition. Perhaps there's a gap in organizational theory here for something like "institutional pareidolia"?
1 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
How would you go about protocolizing this? "Academics are too scared to do ‘career ending’ or controversial research. Let’s encourage people not to be boring & support them. The future of humanity will depend on this to accelerate innovation. We must to be less boring even if it means being wrong! Just be less wrong…" https://x.com/leecronin/status/1863368249083154452
0 reply
0 recast
2 reactions
John Grant pfp
John Grant
@jlg
VibeCheck is a framework for comparing LLMs. It defines a set of "vibes" to measure qualitative differences and provide an alternative perspective from other evaluation metrics that tend to focus on correctness. Aiming to quantify subjective characteristics like humour or formality is clearly reductionist. But vibes are subjective and context-specific. But does reducing vibes to measures, without considering the context, actually help or hinder making meaningful comparisons between LLMs? From a reductivism perspective, VibeCheck also assumes human-AI interactions can be explained from a set of vibes. I think this misses the potential for emergent behaviours and co-evolution of practice. So, is the framework too rigid to capture how human-AI interactions will evolve and adapt? https://arxiv.org/abs/2410.12851
0 reply
0 recast
1 reaction
John Grant pfp
John Grant
@jlg
Strange protocols cometh
0 reply
0 recast
0 reaction
John Grant pfp
John Grant
@jlg
Strange protocols cometh https://arxiv.org/abs/2404.00806
0 reply
0 recast
1 reaction