Alberto Mencarelli
@albm.bsky.social
In search for remote or unexpected associations. Dad, Italian’s Chamber of deputies senior parliamentary official, constitutional law PhD, EU citizen. Currently focused on all things AI
Pinned
Alberto Mencarelli
@albm.bsky.social
· Jul 6
Predictive Parliament: How Next-Token Thinking Can Forge Laws that Actually Work
Large Language Models (LLMs) operate by repeatedly choosing the most probable next word—what engineers call a “token”—based on everything that has come before.
open.substack.com
Returning to the conceptual parallels between LLMs and legislative drafting: how “next-token” logic can sharpen bills, align incentives, and make impact assessments truly measurable. I unpack it all in my latest Substack: open.substack.com/pub/albertom...
With the advent of AI, the cognitive divide is shifting from who has information versus who doesn’t, to who can frame problems for AI models versus who can’t: identifying what matters and crafting effective prompts. Human knowledge work becomes scaffolding for AI cognition.
December 3, 2025 at 8:40 PM
With the advent of AI, the cognitive divide is shifting from who has information versus who doesn’t, to who can frame problems for AI models versus who can’t: identifying what matters and crafting effective prompts. Human knowledge work becomes scaffolding for AI cognition.
What if consciousness — human or artificial — were not something that appears in the world, but the very act through which the world observes itself?
My new Substack post explores this idea: open.substack.com/pub/albertom...
My new Substack post explores this idea: open.substack.com/pub/albertom...
Artificial Intelligence and the Collapse of Meaning
A reflection on the relationship between consciousness, observation, and artificial intelligence
open.substack.com
October 26, 2025 at 3:23 PM
What if consciousness — human or artificial — were not something that appears in the world, but the very act through which the world observes itself?
My new Substack post explores this idea: open.substack.com/pub/albertom...
My new Substack post explores this idea: open.substack.com/pub/albertom...
If AI agents display functional continuity, they should be framed by the law of representation, not treated as substitutes. My latest piece on Substack: open.substack.com/pub/albertom...
October 9, 2025 at 6:58 PM
If AI agents display functional continuity, they should be framed by the law of representation, not treated as substitutes. My latest piece on Substack: open.substack.com/pub/albertom...
The new frontier of computational democracy goes beyond universal access to the power needed for individual AI use.
Allocating finite computational resources means setting priorities and every choice of purpose also means exclusion, a process that must be democratically guided.
Allocating finite computational resources means setting priorities and every choice of purpose also means exclusion, a process that must be democratically guided.
October 2, 2025 at 10:28 AM
The new frontier of computational democracy goes beyond universal access to the power needed for individual AI use.
Allocating finite computational resources means setting priorities and every choice of purpose also means exclusion, a process that must be democratically guided.
Allocating finite computational resources means setting priorities and every choice of purpose also means exclusion, a process that must be democratically guided.
The mirror at the end of the feed open.substack.com/pub/albertom...
The Mirror at the End of the Feed
In its early days the web opened like a map: it turned search into a common skill and widened access to knowledge.
open.substack.com
September 23, 2025 at 9:11 AM
The mirror at the end of the feed open.substack.com/pub/albertom...
The voice mode of next-gen AI devices should act as a distributive multimodal router, invoking other multimodal capabilities whenever needed to enrich the core conversational experience and seamlessly interfacing with legacy devices to deliver content too complex for voice alone.
September 6, 2025 at 11:08 PM
The voice mode of next-gen AI devices should act as a distributive multimodal router, invoking other multimodal capabilities whenever needed to enrich the core conversational experience and seamlessly interfacing with legacy devices to deliver content too complex for voice alone.
Returning to the conceptual parallels between LLMs and legislative drafting: how “next-token” logic can sharpen bills, align incentives, and make impact assessments truly measurable. I unpack it all in my latest Substack: open.substack.com/pub/albertom...
Predictive Parliament: How Next-Token Thinking Can Forge Laws that Actually Work
Large Language Models (LLMs) operate by repeatedly choosing the most probable next word—what engineers call a “token”—based on everything that has come before.
open.substack.com
July 6, 2025 at 2:02 PM
Returning to the conceptual parallels between LLMs and legislative drafting: how “next-token” logic can sharpen bills, align incentives, and make impact assessments truly measurable. I unpack it all in my latest Substack: open.substack.com/pub/albertom...
Exploring whether differential privacy can bridge the GDPR’s data-minimisation rule and the EU AI Act: open.substack.com/pub/albertom...
Individual Intelligence vs. Collective Intelligence: Why Differential Privacy might be the Bridge between GDPR e EU AI Act
Europe stands at a crossroads that will shape our digital future.
open.substack.com
July 1, 2025 at 4:27 PM
Exploring whether differential privacy can bridge the GDPR’s data-minimisation rule and the EU AI Act: open.substack.com/pub/albertom...
Is AI Driving Fertility Change—or Simply Reacting to It? Rethinking the Feedback Loop Between Virtual Companionship and Birthrates open.substack.com/pub/albertom...
Is AI Driving Fertility Change—or Simply Reacting to It? Rethinking the Feedback Loop Between Virtual Companionship and Birthrates
Global fertility has reached its lowest recorded level—about 2.3 births per woman in 2023, half the figure of the early 1950s—and nearly every OECD country now sits below the replacement threshold of ...
open.substack.com
June 28, 2025 at 1:30 PM
Is AI Driving Fertility Change—or Simply Reacting to It? Rethinking the Feedback Loop Between Virtual Companionship and Birthrates open.substack.com/pub/albertom...
Generative AI thrives on pattern.
Human discourse drifts into improvisation. An inverse correlation?
A brief reflection
open.substack.com/pub/albertom...
Human discourse drifts into improvisation. An inverse correlation?
A brief reflection
open.substack.com/pub/albertom...
When Prediction Meets Improvisation
There seems to be a peculiar inverse correlation at play—one that says much about the strange entanglement of human and artificial intelligence in our time.
open.substack.com
June 17, 2025 at 1:42 PM
Generative AI thrives on pattern.
Human discourse drifts into improvisation. An inverse correlation?
A brief reflection
open.substack.com/pub/albertom...
Human discourse drifts into improvisation. An inverse correlation?
A brief reflection
open.substack.com/pub/albertom...
A hidden risk in AI adoption for knowledge work: top senior professionals quietly resist reasoning models, seeing them as threats to their expertise. The challenge isn’t tech—it’s symbolic. AI should be framed as a lever, not a rival.
June 4, 2025 at 10:28 AM
A hidden risk in AI adoption for knowledge work: top senior professionals quietly resist reasoning models, seeing them as threats to their expertise. The challenge isn’t tech—it’s symbolic. AI should be framed as a lever, not a rival.
‘In the beginning was the Word,’ the Gospel of John begins. An idea that may offer insight into how LLMs operate: for them, too, the word brings form to emptiness—not thought preceding speech, but speech generating the appearance of thought. The token precedes the concept. #AI #philosophy
May 29, 2025 at 8:31 AM
‘In the beginning was the Word,’ the Gospel of John begins. An idea that may offer insight into how LLMs operate: for them, too, the word brings form to emptiness—not thought preceding speech, but speech generating the appearance of thought. The token precedes the concept. #AI #philosophy
I expect that AI will ultimately deepen the tension between biological inheritance and digital inheritance, reshaping our concept of personal legacy and shedding new light on global declining fertility.
May 5, 2025 at 2:18 PM
I expect that AI will ultimately deepen the tension between biological inheritance and digital inheritance, reshaping our concept of personal legacy and shedding new light on global declining fertility.
The proper functioning of a democratic system does not require the affirmative adherence of individuals to alternative doctrines, but is sufficiently preserved where individuals are induced to perceive the intrinsic unsustainability or impracticability of their own positions.
April 26, 2025 at 8:20 AM
The proper functioning of a democratic system does not require the affirmative adherence of individuals to alternative doctrines, but is sufficiently preserved where individuals are induced to perceive the intrinsic unsustainability or impracticability of their own positions.
The ultimate political move on the global geopolitical chessboard would be the election of a U.S. cardinal as Pope.
April 25, 2025 at 7:37 AM
The ultimate political move on the global geopolitical chessboard would be the election of a U.S. cardinal as Pope.
A potential emerging risk is the delegation of epistemic responsibility to AI: as artificial intelligence improves in accuracy and fact-checking, humans may grow increasingly complacent in verifying the truth of their own claims, unwittingly turning to AI to fill that role.
April 24, 2025 at 11:18 AM
A potential emerging risk is the delegation of epistemic responsibility to AI: as artificial intelligence improves in accuracy and fact-checking, humans may grow increasingly complacent in verifying the truth of their own claims, unwittingly turning to AI to fill that role.
Improvisation can be a mask—narcissistic, tactical, allergic to limits. True flexibility is slower, reasoned, born of doubt. One avoids the mirror, the other breaks it.
April 19, 2025 at 7:53 PM
Improvisation can be a mask—narcissistic, tactical, allergic to limits. True flexibility is slower, reasoned, born of doubt. One avoids the mirror, the other breaks it.
This infographic, generated via GPT-4o’s native image mode, draws a parallel between AI’s next token prediction and legislative drafting: articles and words act as tokens in a predictive process that absorbs or precludes amendments based on context.
March 26, 2025 at 8:59 AM
This infographic, generated via GPT-4o’s native image mode, draws a parallel between AI’s next token prediction and legislative drafting: articles and words act as tokens in a predictive process that absorbs or precludes amendments based on context.
It cannot be said lightly, but never before has the saying ‘si vis pacem, para bellum’ been more relevant for the EU than it is now.
February 28, 2025 at 8:03 PM
It cannot be said lightly, but never before has the saying ‘si vis pacem, para bellum’ been more relevant for the EU than it is now.
Fewer words, fewer actions. The weight lifts, the essential stays.
Less is more
Less is more
January 3, 2025 at 4:12 PM
Fewer words, fewer actions. The weight lifts, the essential stays.
Less is more
Less is more
disenchantment, as long as it doesn’t turn into cynicism, can be a great way to stop being so hard on yourself
November 20, 2024 at 4:05 PM
disenchantment, as long as it doesn’t turn into cynicism, can be a great way to stop being so hard on yourself
Don’t die
darioamodei.com/machines-of-...
darioamodei.com/machines-of-...
Dario Amodei — Machines of Loving Grace
How AI Could Transform the World for the Better
darioamodei.com
October 12, 2024 at 3:03 PM
Don’t die
darioamodei.com/machines-of-...
darioamodei.com/machines-of-...
If there is one medium to long-term effect of AI that I expect to be particularly disruptive for the world of knowledge, it is the merging of specialized disciplines. Interdisciplinarity will be the common defining trait of future intelligences, both human and artificial.
October 10, 2024 at 2:10 PM
If there is one medium to long-term effect of AI that I expect to be particularly disruptive for the world of knowledge, it is the merging of specialized disciplines. Interdisciplinarity will be the common defining trait of future intelligences, both human and artificial.