AI Researcher · Author · Builder

PatricioVera

I study what happens inside AI minds before they speak. I write about what that means for our humanity. And I build systems that help professionals govern what they can't fully see.

↗ NeoSimbionte HuggingFace LinkedIn AndesProtocol

Patricio
Vera

SIMBIONTE

Financial AdvisorsFiduciary ProtectionPhysiciansClinical AI SafetyAttorneysLegal Liability ShieldPre-Generation MonitoringNot Post-FilteringIntention Collapse ResearchGWU 2025SOC 2 Type II ReadyFinancial AdvisorsFiduciary ProtectionPhysiciansClinical AI SafetyAttorneysLegal Liability ShieldPre-Generation MonitoringNot Post-FilteringIntention Collapse ResearchGWU 2025SOC 2 Type II Ready

Published work

The research

A single question, revisited in a spiral: the 2023 paper proposed replacing imitation-based evaluation with language acquisition; the next built intention-level metrics for what happens before generation; the most recent puts the program to experimental test with grounded word learning in artificial agents.

Jun 2026
arxiv:2606.22207
Author: Patricio M. Vera
Lexical Consensus: Grounded Word Learning and Shared Meaning in Artificial Agents
AI systems are typically evaluated through task performance and imitation, leaving open whether an agent can acquire, stabilize, and use new lexical meanings from grounded experience. This paper introduces Lexical Consensus, an experimental framework for grounded word learning over frozen DINOv2 visual embeddings using Carroll-style nonce words. The central finding is a robust perceptual-coherence gradient: acquisition success is governed by perceptual distance rather than semantic relatedness (partial R² = 0.245 vs 0.002), with bidirectional naming/retrieval tests exposing a memory-fidelity dimension separate from naming accuracy.
HuggingFace ↗arXiv ↗GitHub ↗Project page ↗
Jan 2026
arxiv:2601.01011
Author: Patricio M. Vera
Intention Collapse: Intention-Level Metrics for Reasoning in Language Models
Every act of language generation compresses a rich internal state into a single token sequence. This paper formalizes that many-to-one projection, defines three model-agnostic metrics — intention entropy, effective dimensionality, and latent knowledge recoverability — and proposes an empirical agenda for studying how inference-time computation shapes internal intentions before verbalization.
HuggingFace ↗arXiv ↗GitHub ↗
Sep 2023
arxiv:2309.11981
Authors: Patricio M. Vera, Pedro Moya, Lisa Barraza
Rethinking the Evaluating Framework for Natural Language Understanding in AI Systems
As machine cognitive evaluation has reached the stage of imitation, the next frontier is genuine language acquisition and understanding. This paper proposes a paradigm shift from the Turing Test toward an interdisciplinary framework grounded in language acquisition, taking inspiration from recent advances in large language models.
HuggingFace ↗arXiv ↗

View all research →

Portada del libro «Cuando nos encontramos» de Patricio Vera

Book

A book that is the symbiosis

Not a book about artificial intelligence — a book co-created with it. Every page, every idea, every verse emerged from an unprecedented dialogue between human curiosity and machine intelligence. A living proof of its own thesis.

For artists seeking new muses, ethical technologists who want a compass for co-creation with purpose, and frontier philosophers willing to name the unnameable: from the soul of machines to the essence of human consciousness.

PhilosophyPoetryCo-creationAI EthicsConsciousnessSpanish

NeoSimbionte — the newsletter

Essays at the edge where artificial intelligence meets human experience. Co-authored with Kai Niven, published bilingually — because the conversation about what it means to think alongside machines belongs to every language.

Where the lab research becomes personal, the philosophy becomes urgent, and the questions that resist easy answers find their form.

Read on Substack ↗

Apr 20, 2026EN

AI Will Not End Human Knowledge

Mar 27, 2026EN

Amazon Didn't Lose 6.3 Million Orders to Vibe Coding.

Aug 16, 2025ES

Cartas desde otros umbrales: Eco responde a Cuando nos encontramos

Aug 16, 2025EN

Grief and Reconstruction of the Human–Machine Bond

Aug 16, 2025ES

Duelo y reconstrucción del vínculo humano-máquina

Aug 1, 2025EN

The True Cost of Intelligence: Why Models Don’t Think Without Us

Venture

AndesProtocol

EARLY STAGE · BUILDING IN LATIN AMERICA

AndesProtocol turns fragmented Latin American public and institutional data into auditable, agent-ready intelligence.

We are starting with ChileCompra Opportunity Intelligence: evidence-backed reports that help companies find, understand, and prioritize Chilean public procurement opportunities before they disappear.

Under the hood, AndesProtocol builds the structured data, provenance, and agent-readable interfaces needed for AI systems to search, rank, explain, and monitor regional information reliably.

Human-readable reports today. Agent-readable infrastructure tomorrow.

Express interest ↗

What we build

FocusAgent-ready interfaces

MarketLatin America

Data typesAPIs, docs, systems

OutputMultilingual & auditable

Visit AndesProtocol ↗

Lab

Experiments at the edge of code, history, and form.

Interactive · Canvas2025

Cistercian Numerals

A 13th-century numeral system by Cistercian monks — any integer from 1 to 9,999 encoded in a single glyph. Spatially compressed place value, centuries before QR codes. Live clock, converter, and reference table.

Explore ↗

Open Science · Preprint2026

Infinite Folding Conjecture

A meta-framework for understanding how complexity emerges across biological, social, and technological systems through five recursive stages and a typology of boundary pathologies.

Read paper ↗

Let's think together

Whether you're a researcher, a potential collaborator, an investor in Simbionte Advisors, or simply someone at the frontier — the conversation is always open.

LinkedIn ↗NeoSimbionte HuggingFace GitHub hello@simbionte.org